Broad-Spectrum Antimicrobial Effects of Photocatalysis Using Titanium Dioxide Nanoparticles Are Strongly Potentiated by Addition of Potassium Iodide

Abstract

Photocatalysis describes the excitation of titanium dioxide nanoparticles (a wide-band gap semiconductor) by UVA light to produce reactive oxygen species (ROS) that can destroy many organic molecules. This photocatalysis process is used for environmental remediation, while antimicrobial photocatalysis can kill many classes of microorganisms and can be used to sterilize water and surfaces and possibly to treat infections. Here we show that addition of the nontoxic inorganic salt potassium iodide to TiO₂ (P25) excited by UVA potentiated the killing of Gram-positive bacteria, Gram-negative bacteria, and fungi by up to 6 logs. The microbial killing depended on the concentration of TiO₂, the fluence of UVA light, and the concentration of KI (the best effect was at 100 mM). There was formation of long-lived antimicrobial species (probably hypoiodite and iodine) in the reaction mixture (detected by adding bacteria after light), but short-lived antibacterial reactive species (bacteria present during light) produced more killing. Fluorescent probes for ROS (hydroxyl radical and singlet oxygen) were quenched by iodide. Tri-iodide (which has a peak at 350 nm and a blue product with starch) was produced by TiO₂-UVA-KI but was much reduced when methicillin-resistant Staphylococcus aureus (MRSA) cells were also present. The model tyrosine substrate N-acetyl tyrosine ethyl ester was iodinated in a light dose-dependent manner. We conclude that UVA-excited TiO₂ in the presence of iodide produces reactive iodine intermediates during illumination that kill microbial cells and long-lived oxidized iodine products that kill after light has ended.

INTRODUCTION

Heterogeneous photocatalysis is the use of photoactivated titanium dioxide to produce reactive oxygen species in order to destroy organic pollutants and to kill different classes of microorganisms (1). TiO₂ is a wide-band gap semiconductor, and when a photon of sufficient energy is absorbed, an electron is excited from the valence band to the conduction band, generating a positive hole in the valence band (2). Since energy levels are not available to promote ready recombination of the electron and the hole (as they are in metallic conductors), the electrons and holes survive long enough to carry out reactions. The electrons can reduce oxygen to superoxide, while at the same time the holes can oxidize water to hydroxyl radicals (3). Hydrogen peroxide and singlet oxygen are also produced (4).

Although TiO₂ can occur in several different crystalline forms (anatase, rutile, and brookite), the anatase form is usually employed for applications in photocatalysis (5). Because the process is heterogeneous, TiO₂ nanoparticles that have the maximum surface area/mass ratio are optimum for efficient catalytic activity. The maximum absorption wavelength of the TiO₂ nanoparticle is about 360 nm (6), equivalent to the semiconductor band gap of 3.35 eV (7). The preparation of TiO₂ known as Degussa or Aeroxide (Evonik) P25 is composed of 21-nm-diameter nanoparticles, which are about 75% anatase, 15% rutile, and 10% amorphous, and P25 has been widely used for photocatalysis applications (8).

There have been many different antimicrobial applications whose mechanisms of actions are based on the generation of reactive oxygen species (ROS) (9). Some of these preparations and techniques can be considered biocides or disinfectants, such as hydrogen peroxide, potassium permanganate, peracetic acid, ozone, etc. (10). However, other approaches can be considered more targeted in nature, as they require the combination of two separate moieties in order to generate ROS. This principle applies to photodynamic inactivation (PDI), where a photosensitizing dye is excited to the long-lived triplet state that can undergo energy transfer reactions producing singlet oxygen or alternatively can carry out electron transfer reactions producing hydroxyl radicals (11 – 13). Antimicrobial PDI is usually carried out with the photosensitizing dyes in solution, but heterogeneous dye systems have been employed, when the dyes are attached to solid supports (14) or polymer films (15). This arrangement makes it easier to remove the dye that remains at the completion of the procedure. Antimicrobial photocatalysis is an attractive alternative to PDI especially for environmental applications, because natural sunlight can be employed to activate the TiO₂, and the TiO₂ is not subject to photobleaching, as is often the case with organic dyes, or tetrapyrroles, that are often used as antimicrobial photosensitizers (16).

We recently reported (17) that a PDI procedure employing the widely used photosensitizing dye methylene blue excited by red (660-nm) light was potentiated by addition of the nontoxic inorganic salt potassium iodide. This was a broad-spectrum potentiation effect giving 1 to 2 logs of additional killing against Gram-positive bacteria, Gram-negative bacteria, and fungi. The mechanism involved the production of short-lived reactive iodine species (iodine radicals), as no long-lived antimicrobial activity remained after cessation of illumination. In another recent report (18), we showed that 10 mM sodium bromide could potentiate TiO₂/UVA antimicrobial photocatalysis, producing a 1- to 3-log increase in broad-spectrum antimicrobial killing. In this instance, the mechanism of action appeared to be the photocatalytic production of hypobromite rather than bromine or tribromide.

In the present study, we asked whether antimicrobial photocatalysis mediated by TiO₂ could also be potentiated by addition of potassium iodide and whether the mechanism was via photocatalytic production of iodine/tri-iodide or hypoiodite.

MATERIALS AND METHODS

Chemicals.

Titanium(IV) oxide (TiO₂) anatase P25, potassium iodide (KI), N-acetyl tyrosine ethyl ester, and all other reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated. Hydroxyphenyl fluorescein (HPF) and Singlet Oxygen Sensor Green (SOSG) were purchased from Molecular Probes, Invitrogen, Bedford, MA. TiO₂ and KI stock solutions were prepared in distilled H₂O (dH₂O) prior to use. We have presented the concentration of TiO₂ as millimolar (10 mM = 800 μg/ml) to allow comparison with KI, although the material is actually nanoparticles. We used 50 mM sodium phosphate buffer (PB) at pH 7.4 for all the experiments. All the photocatalysis experiments except ROS probe experiments were carried out in a 24-well plate under magnetic stirring.

Light source.

We used a 365-nm UVA light-emitting diode (LED) light source (Larson Electronics LLC, Kemp, TX). The irradiance for all experiments was fixed at 16 mW/cm² (1 J/cm² delivered in 1 min), measured by a model IL-1700 research radiometer-photometer (International Light, Inc., Newburyport, MA). The emission spectrum was measured with a spectroradiometer (SPR-01; Luzchem Research, Inc., Ottawa, Ontario, Canada) and showed a peak emission at 365 ± 5 nm.

Fluorescent probe assay for generation of specific ROS.

Ninety-six-well clear-bottom black plates were used for fluorescent-probe experiments. SOSG or HPF (final concentration of 5 μM) was added to 0.1 mM or 1 mM TiO₂ with and without addition of 1 or 10 mM KI in a final volume of 200 μl phosphate-buffered saline (PBS) per well. The fluorescence was detected after each aliquot of 0.5 to 1 J/cm² (dose of UVA light), using a fluorescence spectrometer (SpectraMax M5 plate reader; Molecular Devices, Sunnyvale, CA) The excitation and emission settings were used as recommended by the manufacturer of the probes: excitation at 504 nm and emission at 525 nm for SOSG and excitation at 490 nm and emission at 515 nm for HPF.

Iodination of N-acetyl tyrosine ethyl ester.

Sample solutions (total volume, 400 μl) contained TiO₂ (10 mM), KI (100 mM), and N-acetyl-l-tyrosine ethyl ester (10 mM) in PB (pH 7.4, containing 10% methanol) and were irradiated by UVA LED light (365 nm) with magnetic stirring. An aliquot of the solution (50 μl) was removed at different time point (30 min, 60 min, and 120 min) and centrifuged at 4,000 rpm. It was necessary to use relatively large fluences of light in order to get enough product to allow measurement of the peak area. The supernatants were collected for liquid chromatography-mass spectrometry (LC-MS) analysis. The LC-MS analyses were performed on an Agilent 1260 LC system equipped with a triple-quad mass spectrometer. The LC conditions were as follows: column: C₁₈, 2.1 by 50 mm, 1.8 μm; elution gradient, solution A = acetonitrile, solution B = 10 mM ammonium acetate in water, 2% to 100% of A over 6 min with a flow rate of 0.2 ml/min; ionization mode, negative; and injection volume, 5 μl. The mass of the molecular ion of N-acetyl-3-iodo-l-tyrosine ethyl ester was 378 Da, and the retention time was 5.3 min.

Starch iodine reaction and spectroscopy.

In order to detect any production of iodine, we first used the classical formation of a blue-inclusion complex between starch and iodine. One hundred microliters of 0.5% starch solution (Sigma) was added to 100-μl aliquots withdrawn from a suspension of 10 mM TiO₂ and 100 mM KI after different doses of UVA light irradiation. The absorbance of the iodine-starch mixture was measured at 620 nm using a plate reader (SpectraMax M5; Molecular Devices, Sunnyvale, CA).

In order to detect any production of iodine we attempted to detect the presence of tri-iodide ion (I₃⁻), which is quantitatively formed from the reaction between I₂ and I⁻ (19), using spectroscopy to identify the formation of the peak at 352 nm. TiO₂ (10 mM) plus KI (100 mM) with and without 10⁸ Escherichia coli cells was irradiated with fluences as high as 40 J/cm² of UVA (40 min) with stirring. The mixture was centrifuged and the UV-visible absorption spectrum measured. There were detectable peaks at 352 nm (I₃⁻ has a molar absorption coefficient of 26,400 at 352 nm).

Bacterial strains and culture conditions.

We used the community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) strain USA300 LAC (Los Angeles County clone), Escherichia coli K-12 (ATCC, Manassas VA), and the Candida albicans DAY286 reference strain (a gift from Aaron Mitchell, Department of Microbiology, Columbia University, New York, NY). Bacterial cells were grown in brain heart infusion (BHI) medium at 37°C, and Candida cells were grown in yeast extract-peptone-dextrose (YPD) medium at 30°C. Cells were grown overnight to stationary phase and refreshed the next day to mid-log phase (2 h for bacteria and 4 h for Candida). Cells were collected by centrifugation at 3,500 rpm for 5 min and resuspended in phosphate buffer at a density of 10⁸ cells/ml for bacteria or 10⁷ cells/ml for Candida for further experiments. Cell numbers were estimated by measuring the optical density (OD) at 600 nm (OD of 0.6 = 10⁸ cells/ml). To enumerate CFU per milliliter, a 10-μl aliquot of cells was serially diluted 10-fold in PBS to give dilutions of 10⁻¹ to 10⁻⁵ times in addition to the original concentration, and 10-μl aliquots of each dilution were streaked horizontally on square BHI (bacteria) or YPD (Candida) agar plates. Plates were streaked in triplicate and incubated for 18 h at 30°C (Candida) or 37°C (bacteria) in the dark to allow colony formation.

Antimicrobial photocatalysis.

A cell suspension consisting of 10⁸ cells/ml for bacteria or 10⁷ cells/ml for Candida was mixed with 1 mM, 5 mM, or 10 mM TiO₂ in the presence of various concentrations of KI, and then 500 μl of this mixture was transferred to a 24-well plate and illuminated at room temperature using UVA light under magnetic stirring. The incubation time (between addition of compounds to the bacteria and start of the irradiation) was 2 min. The irradiance was fixed at 16 mW/cm² (1 J/cm² delivered in 1 min). No elevation in temperature (<1°C) was found. Cells in control groups (no treatment, TiO₂ in the dark, and TiO₂–100 mM KI in the dark) were incubated in the dark for the same time as the treatment groups (30 min). The control group with UVA light alone had 100 mM KI but no TiO₂. After each dose of UVA light had been delivered, 10-μl aliquots were withdrawn, serially diluted, and streaked on BHI agar plates for bacteria or YPD agar plates for Candida according to the method of Jett et al. (20). CFU were counted after overnight incubation at 37°C for bacteria or 30°C for Candida.

Addition of bacteria after light activation of TiO₂-KI.

To investigate the microbial killing effect of the solution produced after light activation, we added aliquots of illuminated TiO₂-KI suspension to the bacterial cells. The bacterial pellet was collected by centrifuging 400 μl of 10⁸ cells/ml MRSA or E. coli in BHI at 3,500 rpm for 5min. Four hundred microliters of 1 or 10 mM TiO₂ with addition of 10 or 100 mM KI was illuminated with 20 or 40 J/cm² of UVA light with stirring. At the completion of each illumination (5 min), aliquots (100 μl) of the suspension were removed, added to the bacterial pellet, and gently resuspended. After 30 min of incubation, 10-μl aliquots were taken from each group to determine CFU.

In order to test how long the products retained their antibacterial activity, we repeated the above illumination of 5 mM TiO₂ with 10 mM KI using 40 J/cm² UVA light, and after completion of light delivery we removed aliquots at 30, 60, 120, and 180 min and added them to bacteria as described above. This experiment was repeated with much higher concentrations (10 mM TiO₂, 100 mM KI, and 40 J/cm² UVA), and in this case an aliquot was removed after 24 h in addition to the other times.

TEM.

To investigate the killing effect of TiO₂-KI-UVA, the morphology and ultrastructure of the bacterial cells following treatment were examined by transmission electron microscopy (TEM). E. coli and MRSA cells (10⁸ CFU/ml) were added to a final concentration of 1 mM TiO₂ with or without 10 mM KI solution in PB with illumination of 40 J/cm² of UVA light. Cells were spun down (3,500 rpm, 5 min) immediately after treatment, supernatant was removed, and cells were fixed in 2.5% glutaraldehyde and 2% paraformaldehyde and stored overnight at 4°C. After spinning down (1,200 rpm, 10 min) and decanting the fixative, 0.1 M sodium cacodylate buffer (pH 7.2) was added to the pellets. After fixation, hot agar was added to each pellet. The solidified cell pellets were then processed routinely for TEM. The cell pellets were postfixed in 2% OsO₄ in sodium cacodylate, dehydrated in graded alcohol, and embedded in Epon t812 (Tousimis, Rockville, MD). Ultrathin sections were cut on a Reichert-Jung Ultracut E microtome (Vienna, Austria), collected on uncoated 200-mesh copper grids, stained with uranyl acetate and lead citrate, and examined on a Philips CM-10TEM instrument (Eindhoven, Netherlands) at 80 kV.

RESULTS

Addition of potassium iodide potentiates TiO₂ antimicrobial photocatalysis.

When MRSA cells (10⁸ CFU/ml; Gram-positive bacteria) were irradiated with UVA light with stirring in the presence of TiO₂ (80 μg/ml, 1 mM), there was no loss of viability even with fluences as high as 20 J/cm² (Fig. 1A). When 1 mM KI was added to the mixture, there was a small (nonsignificant) increase in the antimicrobial effect. However, when 10 mM KI was added, there was a large increase in bacterial killing (ranging from 2 logs extra at 5 J/cm², to 4 logs extra at 20 J/cm²). When the KI concentration was further increased to 100 mM, there was even more increased killing, consisting of 1 to 2 logs extra at 5 to 10 J/cm², and complete eradication was achieved (>6 logs killing) at 15 to 20 J/cm².

FIG 1.

KI potentiates the killing of Gram-positive MRSA by TiO₂-UVA in a dose-dependent manner. Bacteria (10⁸) cells/ml) were stirred in a suspension of 1 mM (A) or 5 mM (B) TiO₂ containing the specified concentrations of KI while being illuminated with UVA. At various times, aliquots were withdrawn for determination of CFU. Values are means from 3 repetitions, and error bars are standard deviations (SD).

We repeated the above-described experiments with a higher concentration of TiO₂ (5 mM) (Fig. 1B). Even at this concentration, TiO₂ alone did not produce any photocatalytic killing. However, the effects of potentiation by the addition of KI were even more pronounced than those seen with 1 mM TiO₂. Notably, the killing was potentiated (up to 4 logs extra) by only 1 mM KI, with extra killing also being seen with 10 mM KI (5 logs) and 100 mM KI (eradication; >6 logs).

Figure 2A shows the experiment (1 mM TiO₂) repeated with the Gram-negative bacterial species E. coli. It can first be seen that overall E. coli was significantly more susceptible to the iodide-potentiated killing (but not to killing by TiO₂-UVA alone) than MRSA (Fig. 1A). E. coli was eradicated by 1 mM TiO₂–1 mM KI at 20 J/cm², by 1 mM TiO₂–10 mM KI at 15 to 20 J/cm², and by 1 mM TiO₂–100 mM KI at 10 to 20 J/cm². When the TiO₂ concentration was increased to 5 mM, the effect was even more pronounced (Fig. 2B). Although there was still no killing with TiO₂-UVA alone, 5 mM TiO₂ combined with 1 mM or 10 mM KI led to eradication with light doses greater than 5 J/cm². With TiO₂–100 mM KI there was complete eradication with even the lowest dose of light (5 J/cm²), so no points were plotted on the graph.

FIG 2.

KI potentiates the killing of Gram-negative E. coli by TiO₂-UVA in a dose-dependent manner. Bacteria (10⁸ cells/ml) were stirred in a suspension of 1 mM (A) or 5 mM (B) TiO₂ containing the specified concentrations of KI while being illuminated with UVA. At various times, aliquots were withdrawn for determination of CFU. Values are means from 3 repetitions, and error bars are SD.

Figure 3 shows the experiment repeated with the fungal yeast, C albicans. Because we have previously found that the much larger eukaryotic fungal cells are harder to kill than the smaller prokaryotic bacterial cells, we tested the photocatalysis only with 5 mM TiO₂. Again there was no significant killing with TiO₂-UVA alone, but there was some killing (2 logs) with TiO₂–1 mM KI at 20 J/cm². When 10 mM KI was added, there was almost 5 logs of killing at 20 J/cm², and with 100 mM KI there was complete eradication (>5 logs) at fluences from 10 to 20 J/cm².

FIG 3.

KI potentiates the killing of the fungal yeast C. albicans by TiO₂-UVA in a dose-dependent manner. Yeast cells (10⁷ cells/ml) were stirred in a suspension of TiO₂ (10 mM) containing the specified concentrations of KI while being illuminated with UVA. At various times, aliquots were withdrawn for determination of CFU. Values are means from 3 repetitions, and error bars are SD.

TEM.

The TEM micrographs of MRSA are shown in Fig. 4. The electron-dense TiO₂ nanoparticles can be clearly seen surrounding the bacterial cells in Fig. 4B and C. It can be seen that the cells that received the TiO₂-KI-UVA combination and which were significantly killed (Fig. 4C) show a cell membrane that is substantially damaged and is almost absent in parts (inset). However, the cells that received TiO₂ and UVA without addition of KI and were not significantly killed (Fig. 4B) have cell membranes presenting a largely intact appearance.

FIG 4.

Transmission electron microscopy of MRSA cells. (A) Control; (B) 1 mM TiO₂ and 40 J/cm² UVA; (C) 1 mM TiO₂, 10 mM KI, and 40 J/cm² UVA. Insets, higher magnification showing the cell membrane.

In Fig. 5 are the images obtained with E. coli. In the case of TiO₂-KI-UVA with high killing (Fig. 5C), the membrane appears to be “swollen” (inset), and the cell in the center has an electron-lucent area at one end showing leakage of cellular contents.

FIG 5.

Transmission electron microscopy of E. coli cells. (A) Control; (B) 1 mM TiO₂ and 40 J/cm² UVA; (C) 1 mM TiO₂, 10 mM KI, and 40 J/cm² UVA. Insets, higher magnification showing the cell membrane.

Killing after light.

We wished to investigate how much of the synergistic killing was due to production of a relatively long-lived and stable antimicrobial species, so we added microbial cells at different times after completion of light delivery to a stirred mixture of TiO₂ and KI. We added MRSA cells to a stirred suspension of TiO₂ (1 mM, 10 mM, or 20 mM) containing 0 mM, 10 mM, 100 mM, or 200 mM KI that had been treated with increasing fluences of UVA light (up to 40 J/cm²) 5 min after cessation of the light delivery. The extent of bacterial killing depended on all three parameters (TiO₂ concentration, KI concentration, and UVA fluence), with the highest total combination of these parameters giving six logs of killing (Fig. 6A). In order to test whether the photoproduced antibacterial species (tri-iodide or hypoiodite) were stable for the long term (24 h) or had only short-term stability (up to 2 h), we added MRSA cells to a stirred 5 mM TiO₂–10 mM KI suspension that had been illuminated with 40 J/cm2 of UVA at different times after cessation of illumination, during which time the suspension was stored at room temperature in the dark. As can be seen from Fig. 6B, there was a time-dependent loss of bacterial killing power, which had disappeared by 2 h post-UVA light. In order to see if there was a more long-lasting bactericidal species produced at even higher combination values of the three parameters, we used 10 mM TiO₂, 100 mM KI, and 40 J/cm² UVA. In this case, all the MRSA bacteria were killed (zero CFU remaining) at all time points even up to 24 h postlight (data not shown, as no points were available to draw a graph).

FIG 6.

Effects of adding bacteria after light. (A) MRSA cells were added 5 min after cessation of illumination of TiO₂ under the indicated conditions. (B) MRSA cells were added at different times after cessation of illumination.

ROS fluorescent probes.

In order to gain some information on whether the addition of iodide increased the formation of reactive oxygen species (ROS) produced by the illuminated TiO₂ (for instance by acting as an electron donor) or, conversely, whether the ROS produced by the photocatalysis was responsible for oxidizing the iodide to give iodine radicals, iodine, and hypoiodite, we used two fluorescent probes for ROS that we had previously used in photodynamic therapy studies (21, 22) and asked whether their activation would be quenched by addition of iodide. SOSG is relatively specific for singlet oxygen, and HPF is relatively specific for detecting hydroxyl radicals (23). Figure 7A shows there was no quenching of HPF activation produced by 0.1 mM TiO₂ combined with 1 mM KI, but there was significant quenching of HPF activation (>50%) when 1 mM TiO₂ was combined with 10 mM KI. Figure 7B shows that there was significant quenching of SOSG activation (>50%) by both combinations (0.1 mM TiO₂–1 mM KI and 1 mM TiO₂–10 mM KI).

FIG 7.

Activation of ROS-specific fluorescent probes by photoactivated TiO₂. Probe (5 μM) solution containing TiO₂ (0.1 or 1 mM) was illuminated with UVA light (0 to 8 J/cm2) in the presence of KI (1 or 10 mM). Fluorescence was measured in a plate reader after each aliquot of light was applied. (A) HPF; (B) SOSG. Values are means for 6 wells, and error bars are SD.

Mechanistic assays.

We carried out some chemical assays to determine the reaction mechanism that was occurring when TiO₂ was irradiated with UVA in the presence of KI. First we used the blue color produced by complexation of molecular iodine with starch to demonstrate the formation of I₂/I₃⁻. Figure 8A shows a linear relationship between the absorbance at 630 nm (blue starch-iodine inclusion complex) produced when aliquots of a stirred suspension of 1 mM TiO₂ and 100 mM KI was removed after different fluences of UVA were delivered and added to a starch solution.

FIG 8.

Mechanistic chemical assays. (A) Fluence-dependent increase in absorption (620 nm) of the blue iodine-starch inclusion complex by TiO₂-KI-UVA. (B) Tri-iodide absorption (352 nm) generated by TiO₂-KI-UVA is reduced in the presence of MRSA bacterial cells. (C) Fluence-dependent increase in iodination of N-acetyl-tyrosine ethyl ester in the presence of TiO₂-KI-UVA as measured by LC-MS.

We visually observed a pronounced yellow color (attributed to I₂/I₃⁻) developing in stirred suspensions of TiO₂ and KI with increasing fluences of UVA light. This yellow color was noticeably less apparent when bacterial cells were present in the suspension. Figure 8B shows the absorbance spectra of aliquots of these suspensions that had been removed and centrifuged. The presence of bacteria reduced the size of the peak at 350 nm, consistent with the bacteria having been iodinated by reactive iodine radicals, which reduces the formation of free I₂/I₃⁻ in solution. To confirm the ability of the TiO₂-KI-UVA system to iodinate organic molecules, we used the model tyrosine substrate called ethyl ester of N-acetyl-tyrosine (18). Figure 8C shows a light dose-dependent increase in the area under the curve for the iodinated product (N-acetyl-3-iodotyrosine ethyl ester) as measured by LC-MS separation.

DISCUSSION

For the first time we have shown that TiO₂ antimicrobial photocatalysis is strongly potentiated by addition of the nontoxic salt potassium iodide. The magnitude of the effect was surprisingly large, giving up to six logs of additional killing. Under the conditions we employed, the microbial killing obtained by using TiO₂ and UVA light alone was only minimal. It should be pointed out that we used relatively low concentrations of TiO₂ (1 mM or 5 mM) in order to be able to detect potentiation by added iodide. In fact even with 5 mM TiO₂ (without iodide) and the highest light dose, we found less than 1 log of killing. If we had used 10 mM or, even better, 20 mM TiO₂ (instead of 5 mM TiO₂), we would have detected several logs of bacterial killing, depending on the UVA dose. The extent of microbial killing in our studies depended on the concentration of TiO₂, the delivered fluence of UVA light, and critically on the concentration of added KI. However, when iodide was added at 10 mM, there was 5 logs of killing with E. coli, and with 100 mM iodide there was eradication (>6 logs killing) with only 5 J/cm² of UVA. It is not completely clear why it is necessary to have such a relatively high concentration of iodide to see the full potentiation effect. It is often the case that a surprisingly high concentration of one of the reactants is needed in a bimolecular reaction because the lifetime of the other reactant is surprisingly short. This is consistent with the iodide anions reacting with the positive electron holes that are generated during photocatalysis. It could also be the case that the iodide anions were reacting with hydroxyl radicals (also very short-lived) produced during the photocatalysis.

The killing was broad spectrum in nature, with the Gram-negative E. coli being the most susceptible, followed by the Gram-positive MRSA and then the fungus C. albicans. Even the least susceptible C. albicans was eradicated with 10 mM TiO₂, 100 mM KI, and 5 J/cm² UVA, with no appreciable killing seen using photocatalysis without the added KI.

One key question that needed to be answered was whether the potentiation of microbial killing was solely due to oxidized products of iodide anion (molecular iodine, tri-iodide anion, or hypoiodite) that were formed during the reaction as (more or less) stable products or whether short-lived intermediates, such as iodine radicals (I˙) or iodonium cations (I⁺), also made a contribution to microbial killing. We can see from Fig. 1 that for MRSA cells present during the illumination, we obtained 5 logs of killing using 1 mM and 10 mM KI with 15 J/cm² UVA light. For MRSA cells added soon after the cessation of illumination (5 min) (Fig. 4), it was necessary to use 20 mM TiO₂ with 200 mM KI and 10 J/cm² UVA light to obtain equivalent bacterial killing. This is a several-hundredfold-higher total of the multiplied values of the parameters ([TiO₂] × [KI] × [J/cm²]) needed to obtain killing when bacteria are added after the light than when bacteria are present during the light. Therefore, we are fairly confident in concluding that short-lived reactive intermediates (I⁺ or I˙) are responsible for the majority of the extra microbial killing observed. The identity of the oxidized iodine products produced by the reaction can give some information about the identity of the reactive intermediate. To address this issue, we looked at how long the antimicrobial species lasted for (retained their antibacterial activity) after cessation of illumination. We argued that hypoiodite was less stable than I₂/I₃⁻, and we could distinguish between these chemical species by how long their antibacterial activity lasted. Moreover, hypoiodite may be more microbicidal on a molar concentration basis than I₂/I₃⁻. The fact that at moderate total values of the multiplied parameters (5 mM TiO₂, 10 mM KI, and 40 J/cm²) we observed bacterial killing that lasted for 1 h but had disappeared at 3 h, while at high total multiplied parameters (10 mM TiO₂, 100 mM KI, and 40 J/cm²) the postlight killing lasted for as long as 24 h, suggests to us that a mixture of both hypoiodite and I₂/I₃⁻ is formed during the reaction and that hypoiodite is more bactericidal on a molar basis. At higher total doses, there is sufficient long-lived I₂/I₃⁻ formed for the bactericidal effect to last for 24 h, while the (more toxic but more short-lived) hypoiodite has decayed after 1 to 2 h.

The reason why the balance between hypoiodite and I₂/I₃⁻ is important is that the former is produced by a 2-electron oxidation of I⁻, i.e., H₂O + I⁻ → 2e⁻ + 2H⁺ + IO⁻ (E^o_ox = −0.985 V), while I₂ is formed by 2 separate 1-electron oxidation steps, i.e., 2 I⁻ → 2e⁻ + I₂ (E^o_ox = −0.54 V). The reactive intermediate in the first process is I⁻ → 2e⁻ + I⁺, while the reactive intermediate in the second process is I⁻ → e⁻ + I˙.

Jirousek (24) reported that cationic iodine species (I⁺, I₂⁺˙, and I₃⁺) were the effective iodinating agents when iodide was oxidized in the presence of a protein used for radioiodine labeling. The fact that tyrosine was iodinated during the reaction also suggests the involvement of I⁺, as I˙ is considered to be too weakly reactive a radical to have a reasonable rate of tyrosine iodination (25). We have previously reported that addition of KI potentiated the microbial killing caused by antimicrobial photodynamic inactivation (aPDI) using methylene blue (MB) activated by red light (660 nm) both in vitro and in vivo (17). However, although molecular iodine was produced in the reaction, there was no microbicidal effect when bacteria were added after the end of the illumination period. Moreover, we also found that KI potentiated the aPDI killing mediated by a cationic fullerene derivative excited by UVA light (360 nm) or white light (400 to 700 nm), both in vitro and in vivo (26). However, in both cases the degree of potentiation obtained (1 to 3 logs more killing) did not approach the level of extra killing found in the present study (up to 6 logs). This can be explained by assuming that aPDI with MB and red light or with fullerenes, UVA, and white light can carry out only a 1-electron oxidation producing I˙ as the reactive intermediate and I₂/I₃⁻ as the oxidized iodine species. In contrast, photoactivated TiO₂ can carry out both 1-electron and 2-electron oxidation steps, producing both I˙ and I⁺ as reactive intermediates and I₂/I₃⁻ and HOI as oxidized iodine species. I⁺ and HOI have a higher microbial killing effect than I˙ and I₂/I₃⁻ and can iodinate tyrosine, but HOI much less stable than I₂/I₃⁻, as it decomposes to iodate and iodide: 3IO⁻ → IO₃⁻ + 2I⁻.

Our recent study reported that addition of NaBr could potentiate TiO₂-mediated antimicrobial photocatalysis (18). However, the degree of potentiation obtained (1 to 2 logs of additional killing) was again much less than that seen with the present TiO₂-KI-UVA system. The explanation for this probably lies with the fact that the redox potentials for Br⁻ to Br₂ and Br⁻ to BrO⁻ (−0.76 V and −1.07 V, respectively) are higher than those for I⁻ to I₂ and I⁻ to IO⁻ (−0.54 V and −0.985 V, respectively). Moreover, the amount of molecular Br₂ produced in the previous study was much less (undetectable) than the amount of I₂/I₃⁻ produced in the present study.

Another microbicidal system that is known to carry out 2-electron oxidation of iodide, bromide, and chloride to produce hypohalites is based on peroxidase and hydrogen peroxide (27). The exact type of peroxidase enzyme makes some difference to the comparative microbicidal effect using different halide ions, with myeloperoxidase being most active, followed by eosinophil peroxidase, horseradish peroxidase, and lactoperoxidase (28). Hypohalite ions are produced during the reaction, but it was shown that greater microbial killing was obtained if the bacteria were present at the beginning of the reaction and that there was significantly less killing when the bacteria were added 5 min after mixing and no killing at all (for low concentrations) when the preincubation time was 30 min (29).

We used the commercially available P25 titania nanoparticles with an average diameter of 25 nm (30). There have been many innovative approaches using nanotechnology to improve the effectiveness of titania photocatalysis. Researchers are attempting to dope the TiO₂ nanoparticles with platinum (31), nitrogen (32), graphite (33), or other materials in order to shift the activation wavelength away from the UV into the visible range (34). Other groups have fabricated different types of titania nanostructures such as TiO₂ nanotubes (35) and nano-coated thin films for food preservation (36). Silver nanoparticles were combined with TiO₂ to produced a triply enhanced bactericidal effect by nano-Ag alone, increasing the light absorption in the visible region, and by acting as electron traps to promote charge separation of photoinduced electrons (e⁻) and holes (h⁺) (37).

We will need to address the practical applications of the discovery of the strong potentiation of antimicrobial photocatalysis by added iodide. One of the most important applications of antimicrobial photocatalysis is solar-powered water disinfection in the less-developed world. How realistic is it that iodide could be added at a sufficiently high concentration to make a realistic difference in this scenario? Another possible application is in light-activated photocatalysis for treatment of localized infections arising in implanted or indwelling devices used for medical or surgical procedures. There have been proposals to fabricate various medical devices that are highly prone to infection, such as indwelling catheters (38) and artificial joints (39), to carry a TiO₂ nano-structured coating, so that in the event of an infection occurring, the device could be exposed and irradiated with UVA (or visible light in the case of doped materials) to destroy adherent microbial biofilms. This would be considered medically preferable to complete surgical removal and replacement. In this case it could be envisaged that the device would simply need to be sprayed with KI solution before illumination to take advantage of the 6-log potentiation of antibacterial killing observed in vitro. It must be pointed out that further research would need to be carried out to study just how much potentiation of microbial killing could be obtained using KI in biofilm-dwelling cells. It is widely accepted that biofilm-dwelling microbial cells are much more resistant to many types of antimicrobial and disinfectant approaches than planktonic cells (40). The completely nontoxic nature of KI (it is sold in tablets as a health food supplement [41]) would support this application.