Photobactericidal activity activated by thiolated gold nanoclusters at low flux levels of white light

The emergence of antibiotic resistant bacteria is a major threat to the practice of modern medicine. Photobactericidal agents have obtained significant attention as promising candidates to kill bacteria, and they have been extensively studied. However, to obtain photobactericidal activity, an intense white light source or UV-activation is usually required. Here we report a photobactericidal polymer containing crystal violet (CV) and thiolated gold nanocluster ([Au25(Cys)18]) activated at a low flux levels of white light. It was shown that the polymer encapsulated with CV do not have photobactericidal activity under white light illumination of an average 312 lux. However, encapsulation of [Au25(Cys)18] and CV into the polymer activates potent photobactericidal activity. The study of the photobactericidal mechanism shows that additional encapsulation of [Au25(Cys)18] into the CV treated polymer promotes redox reactions through generation of alternative electron transfer pathways, while it reduces photochemical reaction type-ІІ pathways resulting in promotion of hydrogen peroxide (H2O2) production.

T he discovery of antibiotics in the early 20th Century revolutionised the treatment of bacterial infections but after several decades of overuse, the evolution of antibiotic resistant bacteria is a major threat to the practice of modern medicine [1][2][3][4][5][6][7] . Surfaces in healthcare facilities used to be considered not to have an effect on the spread of infection 8 . Recent studies have reported that bacteria-contaminated surfaces make a significant contribution to the incidence of healthcare associated infections [8][9][10][11][12][13] , and that repeated cleaning/disinfection of the contaminated surfaces, which is commonly performed in healthcare facilities, is not always sufficient to remove pathogens from surfaces. For example, 27% of surfaces in rooms were still contaminated by bacteria after four complete cleaning cycles with disinfectant [14][15][16] .
Photobactericidal agents have gained significant attention as promising candidates to kill bacteria and keep surfaces sterile. Toluidine blue O (TBO), methylene blue (MB) and crystal violet (CV) are widely used for surgical and biological stains and wound disinfection, and are known to have photobactericidal activity 17 . When the agents are exposed to a bright visible light source, they are photoexcited and induce reactive oxygen species (ROS), killing bacteria through damage to the cell membrane and DNA [18][19][20][21][22] . Due to their photobactericidal property, and ease of use, the photobactericidal agents were used to treat infections of blood products and oral infections, and are also effective at killing antibiotic resistant bacteria 22 .
Gold clusters consist of a small number of atoms, and its typical size is less than 2 nm 31,32 . A Au cluster has discrete physico-chemical properties compared to larger Au NPs. In contrast to Au NPs, the cluster does not show surface plasmon resonance, and the electronic band structure of the cluster is different from Au NPs, where the band structure of Au NPs is typically continuous whereas Au nanoclusters form a discontinuous band structure with discrete energy levels 33 .
The motivation herein was to see if discrete molecular gold clusters could promote a greater synergistic effect in enhancing the photobactericidal effect of CV dye. To evaluate this, we synthesize [Au 25 (Cys) 18 ] clusters via a scalable microfluidic approach and incorporated them with CV into a polymer surface to test its bactericidal properties. Surprisingly, this cluster promotes bactericidal activity at very low levels of white light. We suggest that the presence of [Au 25 (Cys) 18 ] promotes redox reactions by generating an electron transfer pathway from CV to the cluster, resulting in enhanced H 2 O 2 formation and bactericidal activity.

Results and discussions
To continuously synthesize [Au 25 (Cys) 18 ], a microfluidic segmented flow system that consisted of a tube-in-tube membrane contactor system and coil flow inverter reactor was used (Supplementary Fig. 1). In this system, Au (I) solution containing 3 mM cysteine was mixed with CO-saturated heptane, which was prepared by the tube-in-tube; The mixture passed through a coil flow inverter reactor at 80°C. In the reactor inverter, [Au 25 (Cys) 18 ] was continuously synthesized with residence time of 3 min. The [Au 25 (Cys) 18 ] solution was stored at room temperature overnight. After that, an organic layer, formed by heptane, on the top of the solution was removed. As shown below, this process exclusively formed [Au 25 (Cys) 18 ] clusters with no evidence of larger or smaller clusters.
The UV-Vis spectrum of the [Au 25 (Cys) 18 ] solution was measured from 300 to 900 nm. As shown in Fig. 1a, the nanoclusters gave multiple absorption peaks at 400, 450, 670 and 780 nm which are in agreement with the absorbance features of thiolated Au 25 clusters reported previously 34 . It was confirmed that the concentration of synthesized [Au 25 (Cys) 18 ] was about 0.12 mM (7.2 × 10 16 [Au 25 (Cys) 18 ] clusters mL −1 ). ESI-MS was used to determine the molecular composition of [Au 25 (Cys) 18 ]. As shown in Fig. 1b, it was observed that the most intense set of peaks were at m/z~2361.4 over a range of m/z 1500-5200, and Fig. 1c shows that the base peak at m/ z 2361.4 (peak #1) was accompanied with a group of similar small peaks (peak #2 to #7) corresponding to H + dissociation and Na + coordination to [Au 25 (Cys) 18 ]. Isotope pattern analysis of peak #1 showed that the peak spacing between 12 (Fig. 1d and Supplementary Fig. 2). Additionally, other sets of cluster peaks were observed at m/z~1808.6, 2411, and 3542.6, indicating 2-, 3-and 4-charged gas phase ions. All peaks corresponded to [Au 25 (Cys) 18 ], indicating that no other clusters were formed 35 . TEM (Fig. 1a inset) shows that the size of the [Au 25 (Cys) 18 ] clusters was <2 nm 36 .
A swell-encapsulation-shrink process was employed to produce photobactericidal silicone. Fluorescence microscopy, UV-Vis spectroscopy and XPS confirmed that after the treatment, CV molecules and [Au 25 (Cys) 18 ] clusters penetrated into the polymer matrix ( Supplementary Fig. 3). As shown in Fig. 2a, after [Au 25 (Cys) 18 ] or cysteine treatment, the silicone substrate maintained its colour, while CV treatment changed from white to violet. The colour of the polymer containing CV&[Au 25 (Cys) 18 ] was more intense than the sample with CV alone. Figure 2b shows UV-vis absorbance spectra of the control and treated samples at a wavelength of 400-800 nm. All of the CV-treated polymers have a main absorbance at 595 nm with a shoulder peak at 541 nm. The absorbance of [Au 25 (Cys) 18 ] encapsulated sample was broader and of higher intensity compared to the sample with CV alone. Of materials used for the encapsulation, CV is the only material containing chlorine (Cl). Thus, a change in the weight percentage (wt%) of Cl was determined before and after additional encapsulation of [Au 25 (Cys) 18 ] by X-ray fluorescence spectroscopy (XRF). Cl wt% in CV&[Au 25 (Cys 18 )] polymer was 1.8 times higher than the polymer containing CV alone. This was similar with the increase of light absorbance of the material at 595 nm after additional encapsulation of [Au 25 (Cys) 18 18 ] solution: −31.8 mV) attracts more CV + during the encapsulation process 37 .
Staphylococcus aureus and Escherichia coli are common pathogens in healthcare facilities and can cause a range of infections from local skin and catheter-related infections to sepsis and even death of patients 38,39 . The bactericidal activity of the polymers against S. aureus 8325-4 and E. coli ATCC 25922 was tested in the dark and under white light (average 312 lux), and the result was compared to that of a control comprised of solvent treated silicone. Twenty-five microlitres of the bacterial suspension containing~10 5 CFU mL −1 was inoculated on to each sample surface and then exposed to the white light source or maintained in a dark room at 20°C. Supplementary Fig. 5 shows the distribution of white light intensity. The intensity ranged from 200 to 429 lux (0.03 to 0.06 mW cm −2 ) and the light emission wavelength mainly ranged from 400 to 800 nm 40 . Figure 3a, b shows the bactericidal activity of the samples against S. aureus after 6 h incubation in the dark and in white light. After 6 h incubation in the dark, compared to the control, a reduction in the number of viable bacteria was not observed on either silicone samples containing only CV or [Au 25 (Cys) 18 ] alone, whereas a statistically significant reduction in bacterial numbers was observed on the silicone samples containing cysteine only (0.17 log reduction, T-test: P < 0.05) and CV& [Au 25 (Cys) 18 ] (0.25 log reduction, T-test: P < 0.1). Upon 6 h exposure to white light, a reduction in the number of viable bacteria was not observed on samples containing CV only or [Au 25 (Cys) 18 ] only, compared to the control. Bactericidal activity of silicone containing cysteine alone was observed after 6 h exposure to white light, but the activity was not enhanced compared to that in the dark (T-test: P > 0.1). The bactericidal activity of the combined CV&[Au 25 (Cys) 18 ] encapsulated sample was significantly enhanced after 6 h exposure of white light; the number of viable bacteria on the sample with CV&[Au 25 (Cys) 18 ] fell to below the detection (<10 2 CFU mL −1 ), indicating a >3.3log reduction in viable bacteria.  E. coli was more resistant than S. aureus requiring a longer exposure time of white light to achieve a significant reduction in the number of viable bacteria. We attribute this to differences in the cell wall, where E. coli-a Gram-negative bacterium-contains a double membrane structure compared to S. aureus-a Gram-positive bacterium-which contains only a single membrane barrier 41 . The outer membrane of Gram-negative bacterium decreases molecular penetration, and is often responsible for increased resistance to antibacterial agents 41,42 .
It is known that the photobactericidal mechanism of CV starts with the absorption of light, resulting in an excited singlet state which has paired electron spin 22,28 . This either returns to the ground state or transforms to a triplet state, which has unpaired electron spin 22,28 . Molecules in the triplet state undergo two photochemical reaction processes known as Type-І and -ІІ. For Type-І, redox reactions occur with their environment resulting in ROS generation 22,28 . For Type-ІІ, the energy is transferred to ground state molecular oxygen, resulting in the transformation from triplet oxygen ( 18 ] of E. coli in white light (n=6 independent samples). Bacteria inoculated materials were exposed to white light with an intensity from 200 to 429 lux. Data presented as mean ± SD. a CFU mL −1 indicates Colony forming unit per mL. Blue star represents below detection limit: <10 2 CFU mL −1 .
death 26 . The polymer containing CV only did not show any photobactericidal activity herein, upon exposure to low intensity white light. However, the addition of [Au 25 (Cys) 18 ] to CV resulted in potent photobactericidal activity under identical conditions. To determine the species responsible for the bactericidal effect observed in our CV&[Au 25 (Cys) 18 ] system, ROS scavenger/ quencher assays and 1 O 2 phosphorescence measurements were carried out. In the ROS scavenger/quencher assay, photobactericidal activity was measured in the presence of scavengers; superoxide dismutase (SOD) for superoxide radicals (O 2 − ), catalase for hydrogen peroxide (H 2 O 2 ), mannitol for hydroxyl radicals ( • OH) and L-histidine for singlet oxygen species ( 1 O 2 ) 29,43-45 . As shown Fig. 4a, the polymer with CV&[Au 25 (Cys) 18 ] exhibited potent photobactericidal activity against S. aureus with >3.3-log reduction compared to the control after 6 h exposure of 312 lux white light without addition of ROS scavenger/quencher. When Lhistidine ( 1 O 2 quencher) was added, the photobactericidal activity of the polymer was slightly reduced, but retained strong bactericidal activity with a 2.9 log reduction in the number of viable bacteria. O 2 − and • OH scavenging by SOD and mannitol respectively did not cause any significant change in photobactericidal activity. However, the photobactericidal activity decreased from >3. 3 18 ] to CV appears to attenuate the Type-ІІ pathway.
To understand the photoreaction mechanism of the polymer sample containing CV&[Au 25 (Cys) 18 ], photocurrent measurements, steady state and time-resolved photoluminescence (PL) spectroscopies were employed. Transient photocurrent responses of CV only and CV&[Au 25 (Cys) 18 ] treated polymers were measured under several on-off cycles of white light irradiation. As shown in Fig. 5a, a greater rise in photocurrent of the polymers containing CV&[Au 25 (Cys) 18 ] compared to CV alone was observed, indicating a higher CV concentration and/or a higher separation efficiency of electron-hole pairs. Figure 5b shows PL spectra of the samples with CV only and CV&[Au 25 (Cys) 18 ] from 600 to 800 nm. The intensity of the PL spectrum is a direct measurement on recombination rate of electron-hole pair. The higher the peak intensity of the spectrum, the easier the recombination of the electron-hole pair. A PL peak of the polymer sample with CV only was observed at~675 nm, and the PL peak of the sample with CV&[Au 25 (Cys) 18 ] was lower than that of CV only indicating that the recombination rate of the photogenerated electron-hole pairs on CV&[Au 25 (Cys) 18 ] is slower than that of CV only. This indicates that additional encapsulation of [Au 25 (Cys) 18 ] into CV-treated polymer improves the photoelectrochemical properties of the sample. Figure 5c shows 18 ] on the CV-treated surface acts as an electron acceptor, resulting in better electron-hole pair separation, a reduction in recombination, and enhanced photobactericidal properties 48,49 .
In order to determine the band offset of the CV and [Au 25 (Cys) 18 ], XPS and UV-Vis spectroscopy were used, with the results shown in Fig. 6. For XPS analysis, the adventitious C 1s peak at 284. 8 Fig. 6b, e). A difference of binding energy between N 1s core level (CL) and homo band was investigated in CV only and [Au 25 (Cys) 18 ] only samples (Supplementary Fig. 6c, f). The same functions were used to peak fit N 1s peaks in CV&[Au 25 (Cys) 18 ] to determine the binding energy (Fig. 6a). The N 1s CL environments can be considered equivalent in these materials, and can therefore be used as a reference point to determine the band alignment. As shown in Fig. 6b an adverse effect on hospital staff and patients. Secondly, because light sources typically range from 300 (corridors, rooms) to 90,000 lux (operating theatre) in hospitals and healthcare facilities (Table 1), potent photobactericidal activity should be present at various lighting conditions. Titanium dioxide (TiO 2 ) and zinc oxide (ZnO) nanoparticles are the most extensively studied photocatalysts, and they exhibit a broad photobactericidal spectrum [52][53][54] . However, their bactericidal activity is negligible under white light sources, which are commonly used in healthcare facilities, because TiO 2 and ZnO nanoparticles are UV-active photocatalysts, and the portion of UV light present in indoor lighting is typically extremely low 52 . To produce enhanced indoor photocatalysts, TiO 2 based nanocomposites using Ag NPs and Au NPs, carbon nanotube (CNT), and graphene oxide (GO) etc have been studied [53][54][55] . Although some of the composites exhibited bactericidal activity under a white light source, such studies typically employed an intense light source of >1300 lux (0.2 mW cm −2 ) indicating lack of feasibility for use in healthcare facilities (  18 ] displayed potent bactericidal activity against S. aureus and E. coli, which are both associated with hospital acquired infections (HAIs). An illumination with a white light source (312 lux) of similar intensity to that commonly found in healthcare facilities resulted in a significant enhancement of the bactericidal activity of the polymer with >3.3 and 2.8 log reductions for S. aureus and E. coli respectively after 6 and 24 h of exposure to white light. Our studies of the photobactericidal mechanism showed that upon white light illumination, photogenerated electrons in CV likely flow into [Au 25 (Cys) 18 ] (Fig. 6). This indicated that redox reaction was induced from [Au 25 (Cys) 18 ]. Chemical scavenger and TRNIR spectroscopy studies showed that the Type-II pathway for forming 1 O 2 was suppressed in the CV&[Au 25 (Cys) 18 ] system, and that the redox reaction to form H 2 O 2 was promoted. The photobactericidal polymer containing CV&[Au 25 (Cys) 18 ], which we report herein is a promising candidate for use in healthcare environments to prevent the spread of HAIs because its potent photobactericidal activity at ambient light levels (Table 1).

Methods
Synthesis of gold nanocluster ([Au 25 (Cys) 18 ]). Supplementary Fig. 1 shows a set-up of microfluidic segmented flow system to synthesize [Au 25 (Cys) 18 ]. It consists of a tube-in-tube membrane contactor, and a coil flow inverter reactor. The tube-in-tube contactor consists of polytetrafluoroethylene (PTFE) tubing (inner diameter: 3.2 mm, VICI Jour), and Teflon AF-2400 tubing (inner diameter: 0.8 mm, Cambridge Reactor Design Ltd) located in PTFE tubing. CO (BOC) was dissolved into heptane in the tube-in-tube contactor and then the mixture was saturated under 500 kPa. 21.8 mg of cysteine and 3 mL of tetrachloauric acid stock solution (59.5 mM, Sigma-Aldrich) were mixed in de-ionized (DI) water to form 40 mL of the thiolate-Au (I) complex solution (3 mM), the pH of the solution was maintained at 11.6 using sodium hydroxide (NaOH, 2 N, Sigma-Aldrich). The complex was mixed with CO-saturated heptane and then it passed through the coiled flow inverter at 80°C. The coiled flow inverter reactor consists of 100 coils of fluorinated ethylene propylene (FEP) tubing (inner diameter: 1 mm, VICI Jour). The diameter of the coil was about 1 cm and 90°bend was applied at every five coils. After the reactor, the solution was mixed with nitrogen gas in the collection flask. Through the microfluidic segmented flow system, [Au 25 (Cys) 18  Electrospray ionization-mass spectrometry. To characterise [Au 25 (Cys) 18 ], the dried nanoclusters were mixed with DI water containg 0.01 µM cesium acetate, and then the solution was infused into an electrospray ionization-mass spectrometer (ESI-MS, Q-TOF 6510, Agilent Technology) at a flow rate of 20 µL min −1 . The electrospray ionization (ESI) was operated in negative mode, and heated nitrogen (the drying gas) was supplied into the spectrometer at a flow rate of 5 L min −1 . The negatively charged ions were analysed by mass spectrometer (MS).
where C and A indicate the concentration (mM) of [Au 25 (Cys) 18 ] clusters and the absorbance of the cluster solution at 670 nm, respectively, ε and ℓ represent molar absorption coefficient at 670 nm and path length through the sample, respectively.
Preparation of photobactericidal surface. In a mixture of acetone (140 mL), [Au 25 (Cys) 18 ] clusters (28 mL), distilled (DI) water (112 mL) or acetone (140 mL) and DI water (140 mL), 224 mg of crystal violet (CV, Sigma-Aldrich) was dissolved at a concentration of 800 ppm. Silicone sheet (4.5 cm × 7.0 cm) was immersed in 280 mL of was dissolved for 24 h. The sheet was collected from the solution, washed by DI water two times to remove unbound materials, and then air dried for 24 h in a dark room. After that, the sheet was cut into small pieces (1.5 cm × 1.5 cm).
Characterisation of photobactericidal surface. UV-Vis absorbance spectra of the control, and treated silicone samples were measured by UV-Vis spectrometer (Lambda 25, PerkinElmer Inc.), and the absorbance was measured in a range of 400 −1000 nm.
To determine the CV diffusion in silicone, the polymer was immersed in CV solution for 0.5, 1, 5 and 24 h, and it was sliced into 30 μm thickness using a microtome. The sliced sample was mounted onto a glass slide and then its side section was imaged using fluorescence microscopy (Olympus BX63, Tokyo, Japan) with TRITC filter (excitation/emission wavelengths 545 nm/620 nm). The samples were also photographed by optical microscopy, and then the image was analysed using ImageJ (http://imagej.nih.gov/ij/) to obtain the cross-sectional intensity profiles.
X-ray photoelectron spectroscopy (XPS). XPS depth profile was employed to determine the existence of [Au 25 (Cys) 18 ] within the treated polymer matrix and at the polymer surface. The binding energies of Au atom were determined after 0, 200 and 400 s of sputtering.
X-ray fluorescence (XRF) analysis. XRF analysis was employed to determine a change of CV concentration after additional encapsulation of [Au 25 (Cys) 18 ] into CV-treated polymer. The treated polymer (1.5 cm × 1.5 cm) was placed into pot with colourless substrate and then it was located into XRF spectrometer (15watt Epsilon 4, Malvern Panalytical Ltd, UK).
Bactericidal test. The bactericidal activity of control and the treated polymers were tested against Staphylococcus aureus 8325-4 and Escherichia coli ATCC 25922 under dark and white light conditions. S. aureus and E. coli were stored in brain heart infusion broth (BHI, Oxoid Ltd., Hampshire, England, UK) with 20% glycerol at −70°C, respectively and then they were propagated on either mannitol salt agar (Oxoid Ltd.) or MacConkey agar (Oxoid Ltd.) at 37°C, respectively. One of the bacterial colonies grown on the agar was inoculated into BHI broth and cultured at 37°C with 200 rpm shaking. After 18 h of culture, the bacteria were harvested through centrifugation (20°C, 2795 × g for 5 min), washed using 10 mL of PBS to remove BHI broth and then centrifuged to get the bacteria resuspended into 10 mL of PBS. The washed bacterial suspension was 10000-fold diluted to obtain~10 5 CFU mL −1 . As shown in Supplementary Fig. 7, 25 μL of the bacterial suspension was inoculated onto the control and the treated sample surface and then the samples were located in a colourless petri dish containing wet filter paper to obtain constant humidity. After that, the samples were exposed to white light (Osram L58W/865Lumilux, Munich, German) while a same set of samples was kept in a dark room, and then the samples were placed in 450 μL of PBS and vortexed for 1 min to wash out bacteria from the sample into the PBS. After a serial dilution, S. aureus and E. coli suspensions (each 100 μL) were plated onto mannitol salt agar and MaConkey agar, respectively and cultured at 37°C for 24 h. The bacterial colonies which were grown in the agar were counted. Each experiment contained two technical replicates and the experiment was reproduced three times · Detection of superoxide radical, hydrogen peroxide, hydroxyl radical and singlet oxygen species. S. aureus 8425-4 and catalase (Sigma-Aldrich), L-histidine (Sigma-Aldrich), mannitol (Sigma-Aldrich) and superoxide dismutase (SOD, Sigma-Aldrich) which are scavengers and quenchers of ROS were used. Catalase was used at a concentration of 6-14 unit mL −1 in bacterial solution to remove hydrogen peroxide (H 2 O 2 ). SOD was used at a concentration of~20 unit mL −1 in bacterial suspension to remove superoxide (O 2 − ) radicals. Mannitol was used at a concentration of 82 mM in bacterial solution to eliminate hydroxyl radicals ( • OH). L-histidine was used at a concentration of 2 mM in bacterial solution as singlet oxygen ( 1 O 2 ) quencher. The catalase and SOD scavengers which are enzyme act extracellularly since they are large molecules. Thus, the catalase intercepts the hydrogen peroxide coming from the surface. Twenty-five microlitres of the bacterial suspension containing ROS scavenger or quencher was inoculated onto antimicrobial sample, the samples were located on to a colourless petri dish containing wet filter paper, and then they were was exposed to white light for 6 h. After that, samples were located in 450 μL of PBS and vortexed for 1 min to wash out bacteria from the sample to PBS. After a serial dilution, 100 μL of bacterial suspension was plated onto mannitol salt agar and cultured at 37°C for 24 h. The bacterial colonies which were grown in the agar were counted. Each experiment contained two technical replicates and the experiment was reproduced three times · Detection of time-resolved singlet oxygen phosphorescence. In order to detect singlet oxygen ( 1 O 2 ) phosphorescence, a near-infrared sensitive thermoelectrically cooled photomultiplier was employed. The silicone sample was placed onto a slide glass and irradiated by Nd:YAG laser operating at 532 nm. A PC-mounted multiscaler board with pre-amplifier (MSA-300, Becker-Hickl) was used as the photon counting system to measure singlet oxygen phosphorescence at a wavelength of~1270 nm. After the measurement, the data was analysed using FluoFit software (PicoQuant GmbH).
Photoelectrochemical measurement and PL spectroscopy. Photocurrent measurements were performed using Metrohm Autolab (PGSTAT302N, Utrecht, Netherlands) with a three-electrode system with an external source of white light. Ag/AgCl reference electrode, and platinum sheet as a counter electrode. The photobactericidal polymer (CV only and CV&[Au 25 (Cys 18 )]) served as the working electrode. The electrolyte used in the system was 0.1 M Na 2 SO 4 . PL spectra of samples were measured in a wavelength of 600 to 800 nm using steady state PL spectrometer (FluoroMax, Horiba Scientific, Kyoto, Japan). The excitation wavelength was~574 nm.
Zeta potential measurement. Zeta potential of the [Au 25 (Cys) 18 ] in the mixture of water and acetone were measured by Delsa zeta potential anlyzer (DelsaMax-Pro, Beckman Coulter) with the available flow cell system in batch mode at 22°C.
Measurement of band offset. XPS was employed to determine the band offset of CV and [Au 25 (Cys) 18 ]. The energy difference between a CL and homo band maximum (HBM) for individual materials was measured and then the difference between CLs of CV and [Au 25 (Cys) 18 ] was measured. Subsequently, the band alignment and energy offset were determined by Eq. (2)