Boosting Membrane Interactions and Antimicrobial Eﬀects of Photocatalytic Titanium Dioxide Nanoparticles by Peptide Coating

Photocatalytic nanoparticles oﬀer antimicrobial eﬀects under illumination due to the formation of reactive oxygen species (ROS), capable of degrading bacterial membranes. ROS may, however, also degrade human cell membranes and trigger toxicity. Since antimicrobial peptides (AMPs) may display excellent selectivity between human cells and bacteria, these may oﬀer opportunities to eﬀectively “target” nanoparticles to bacterial membranes for increased selectivity. Investigating this, photocatalytic TiO 2 nanoparticles (NPs) are coated with the AMP LL-37, and ROS generation is found by C 11 -BODIPY to be essentially unaﬀected after AMP coating. Furthermore, peptide-coated TiO 2 NPs retain their positive 𝜻 -potential also after 1–2 h of UV illumination, showing peptide degradation to be suﬃciently limited to allow peptide-mediated targeting. In line with this, quartz crystal microbalance measurements show peptide coating to promote membrane binding of TiO 2 NPs, particularly so for bacteria-like anionic and cholesterol-void membranes. As a result, membrane degradation during illumination is strongly promoted for such membranes, but not so for mammalian-like membranes. The mechanisms of these eﬀects are elucidated by neutron reﬂectometry. Analogously, LL-37 coating promoted membrane rupture by TiO 2 NPs for Gram-negative and Gram-positive bacteria, but not for human monocytes. These ﬁndings demonstrate that AMP coating may selectively boost the antimicrobial eﬀects of photocatalytic NPs.


Introduction
Due to growing antibiotic resistance, there is a need for new types of antimicrobial agents. [1,2]5] Among these, photocatalytic nanomaterials display potent antimicrobial effects under illumination.In photocatalysis, free electrons and electron holes are formed on light excitation.8] There are many types of photocatalytic nanomaterials, including TiO 2 , and quantum dots, as well as various carbonbased photocatalysts, such as fullerenes and graphitic carbon nitride nanosheets. [9]mong these, TiO 2 NPs have received particularly intense interest.Photocatalytic effects of TiO 2 NPs have been found to depend, e.g., on crystal structure and morphology. [10]In addition, photocatalytic effects of TiO 2 can be boosted through doping and heterojunction formation, aiming at i) decreasing the band gap (making the nanoparticles excitable by visible light), ii) increasing free radical formation, and iii) decreasing the rate of charge re-combination.Investigating antimicrobial effects of such systems (further examples can be found in), [9,10] Arora et al. found potent antimicrobial effects of TiO 2 NPs against multidrug resistant (MDR) P. aeruginosa under UV illumination. [11]imilarly, Ahmed et al. investigated antimicrobial effects of TiO 2 NPs on 25 P. aeruginosa isolates that were highly or completely resistant to antibiotics alone and found antimicrobial effects of TiO 2 NPs for all MDR isolates.Furthermore, TiO 2 NPs combined with the antibiotics ciprofloxacin, cefepime, amikacin, or ceftriaxone showed synergistic activity against all MDR isolates. [12]Illustrating the role of TiO 2 structure on such effects, Molina-Reyes et al. compared ROS generation and antimicrobial effects of TiO 2 nanoparticles and nanotubes against E. coli.While ROS generation was comparable for these systems, the antibacterial activity was higher for TiO 2 nanotubes, an effect of the latter allowing closer surface contact with bacteria. [13]espite potent antimicrobial effects thus displayed by TiO 2 and other photocatalytic nanomaterials, key challenges in their development toward antimicrobial therapeutics include their toxicity, as well as a currently poor understanding of their antimicrobial spectrum width.Illustrating the latter, Sułek et al. found a 7-log reduction of Gram-positive Staphylococcus aureus (S. aureus) under UV illumination in the presence of porphyrindoped TiO 2 NPs, but a much poorer effect against Gram-negative Escherichia coli (E.coli). [14]In contrast, Gao et al. found Fe 3+doped g-C 3 N 4 to be equally potent against Gram-negative Pseudomonas aeruginosa (P.aeruginosa), Gram-negative E. coli, and Gram-positive S. aureus, [15] whereas Yadav et al. found antimicrobial effects of g-C 3 N 4 nanosheets to be stronger for E. coli than for S. aureus. [16]Considering this scatter in functional performance, further work is clearly needed to clarify how antimicrobial effects and toxicity of photocatalytic nanomaterials depend on differences in membrane structure and composition.Compositionwise, cholesterol is a key component (up to 45% mol) in human cell plasma membranes, but it is not present in bacteria. [17,18][20] Such systems are therefore frequently used as model systems for cell and bacteria membranes, respectively. [3,9]ue to its large band gap, [21] TiO 2 requires illumination in the UV spectrum to display photocatalytic properties.While this may be a disadvantage in practical applications due to limited tissue penetration of UV light, it makes TiO 2 NPs suitable for mechanistic studies, since "in darkness" measurements can be achieved under normal laboratory lighting, conveniently "turning on" ROS generation by UV illumination.Previously, we reported on the influence of lipid composition on membrane susceptibility to photocatalytic degradation by bare TiO 2 NPs.Whereas cholesterol was found to provide stabilization against TiO 2 -mediated oxidation of phosphatidylcholine-based bilayers, anionic phosphatidylglycerol loading increased susceptibility to oxidative destabilization. [22,23]Although such compositional differences thus provide some selectivity between membranes mimicking those of bacteria and human cells, the development of photocatalytic nanomaterials toward antimicrobial therapeutics requires further strengthening of such selectivity.In this context, antimicrobial peptides (AMPs) provide interesting opportunities.AMPs play a key role in innate immunity, and reach antimicrobial potency by several action mechanisms, the most important of which being the physical destabilization of bacterial membranes. [24,25]In addition to such direct destabilization of bacterial membranes, some AMPs interact with bacterial lipopolysaccharides to suppress inflammation and subsequent uncontrolled host response to the infection. [26,27]A key aspect of AMPs is that these can be broadly potent against bacteria, including those resistant to antibiotics, yet display low toxicity against human cells. [28,29]Considering this, AMPs may potentially be used as a surface coating to effectively "target" nanomaterials to bacterial cells.In this study, we therefore investigate if TiO 2 NPs can be coated with AMPs to boost selectivity between bacteria and cell membranes further.In doing so, however, several key questions must be addressed: i) Considering that ROS generation depends on photogenerated electrons and holes reaching the TiO 2 /water interface to react with water and dissolved oxygen to form ROS, how does a (potentially dense) peptide layer influence ROS formation?ii) Since also peptides are susceptible to photocatalytic oxidation, can a window be found, in which the adsorbed peptide layer does not impair ROS formation to allow oxidative degradation of bacterial membranes, yet remains sufficiently stable for allowing targeting of these by the peptide-coated nanoparticles?iii) Since membrane lysis of AMPs depends on their ability to orient and insert into lipid membranes, properties which may be suppressed when AMPs bind to the nanoparticle surface, how do these effects influence the destabilization of lipid membranes?
In addressing these questions, we employed a battery of methodologies previously established by us for investigating photocatalytic degradation of phospholipid membranes by bare TiO 2 NPs, including neutron reflectometry (NR), quartz crystal microbalance with dissipation monitoring (QCM-d), light scattering, and fluorescence spectroscopy.Based on considerations of bacteria and cell membrane composition, we selected model bilayers based on PC, supplemented with either PG (+PG) or cholesterol (+Chol) to mimic bacterial or mammalian membranes, respectively.PC lipid membranes, devoid of PG or cholesterol, were also included in the study as reference samples.Regarding the choice of specific phospholipid acyl chains, palmitoyloleoyl PC and PG were selected for relevance in bacterial and cell membranes, but also for forming largely defect-free supported bilayers, necessary for neutron reflectometry experiments.To understand how the physicochemical results thus obtained for model lipid membranes translate to antimicrobial effects and cell toxicity, in vitro effects of AMP-coated TiO 2 NPs on bacteria and cells were also studied by confocal microscopy and cell toxicity assays.Since photocatalytic nanoparticles are particularly interesting for combatting superficial infections, these studies were performed at slightly acidic and neutral conditions, relevant for the pH range of skin. [30,31]

Dry Lipid Film and Liposome Preparation
Lipid films and liposomes were prepared as described previously. [22]In short, lipids were dissolved in either chloroform or chloroform/methanol (3:1 v/v mixture for dissolving POPG) to form the following lipid mixture: 75/25 POPC/PAPC ("PC"), 50/25/25 POPC/PAPC/POPG ("+PG"), or 65/25/10 POPC/PAPC/Chol ("+Chol") mixtures (molar ratios).The final lipid concentration was 10 mg mL −1 .Note that the inclusion of PAPC was not critically important for oxidative degradation, but still provides an advantage by reducing the UV exposure dose required, and hence also the risk for sample heating.The solvent was then thoroughly removed, first under N 2 flow and then under vacuum overnight.The lipid films were re-hydrated to 1 mg mL −1 with MQ water for liposomes for supported lipid bilayers (SLB), and with buffer for liposomes to be investigated in suspension.The solution was vortexed and bath-sonicated for 30 min to obtain multilamellar vesicles (MLVs).Small unilamellar vesicles (SUVs) for supported bilayers were formed by 15 min of tip sonication (UP50H, Hielscher Ultrasonics GmbH, Germany), which gave SUVs of ≈30 nm diameter.Large unilamellar vesicles (LUVs) were formed by vesicles extrusion through polycarbonate filters (100 nm pores) using a LipoFast miniextruder (Avestin, Ottawa, Canada), to be used for ROS generation studies.

Size and 𝜻-Potential Measurements
Dynamic and electrophoretic light scattering (DLS and ELS; 173°b ack-scattering angle) was performed using a Zetasizer Nano ZSP (Malvern Pananalytical Ltd., Malvern, UK) to obtain particle size and -potential characteristics.Measurements consisted of 10 runs of 10 s each, using automatic attenuation, are reported as the number-average of the effective particle diameters, and were performed in triplicate at 25 °C.Apart from bare and LL-37-coated TiO 2 NPs, such measurements were performed also for LL-37-coated non-photocatalytic mesoporous silica nanoparticles (160±1 nm diameter) [35] to investigate the effects of UV exposure as such on peptide oxidation (Figure S3, Supporting Information).

Small Angle X-Ray Scattering (SAXS)
Structural features of bare TiO 2 NPs were studied using the laboratory-based Xeuss 3.0 instrument (Xenocs, France) at The Center for Scattering Methods (CSM) in the Faculty of Science, Lund University (Sweden), equipped with an X-ray source producing a photon beam with a wavelength of 1.34 Å.The scattering patterns were recorded with an Eiger2 R 1 M 2 D-detector (Dectris) and azimuthally integrated using the XSACT program available with the equipment creating 1D scattering curves.The radially averaged intensity, I(q), was given as a function of the scattering vector  = 4sin/, where  is the wavelength and 2 was the scattering angle.For the measurements, a dispersion of 5000 ppm bare TiO 2 NPs in 10 mM Acetate buffer, pH 3.4, was sealed in a glass capillary and the scattering profile was acquired at 25 °C.

Cryogenic Transmission Electron Microscopy (cryo-TEM)
Cryo-TEM experiments were performed on a JEM-2200FS transmission electron microscope (JEOL) at the National Center for High-Resolution Electron Microscopy (nCHREM), Lund University.Zero-loss images were recorded at 200 kV on a bottommounted TemCam-F416 camera (TVIPS) using SerialEM under low-dose conditions.Specimens were prepared using an automatic plunge system (Leica EM GP) with the environmental chamber at 20 °C and 90% relative humidity.Four mL droplets were deposited on a carbon-coated grid (Ted Pella) and blotted with filter paper to remove excess fluid.The grid was plunged into liquid ethane (−184 °C) to ensure rapid vitrification.Specimens were then stored in liquid nitrogen (−196 °C) until transferred into the microscope using a cryo transfer tomography holder (Fischione Model 2550).Cryo-TEM images for 5000 ppm bare TiO2 NPs in Acetate buffer, pH 3.4, were acquired at different magnifications, and particle size histograms were obtained from cryo-TEM images using ImageJ (National Institutes of Health, Bethesda, USA). [36,37]Weibull fits [38] of particle size histograms were performed to obtain average particle size and standard deviation.

C 11 -BODIPY Oxidation Assay
C 11 -BODIPY experiments were performed as described in detail elsewhere. [22]In brief, 0.5 mol% of the probe was added before lipid film drying under Ar atmosphere.After hydration and extrusion, 0.5 mg mL −1 LUVs were illuminated by UV in situ (254 nm; 3 mW cm −2 ) in the presence or absence of either bare or peptide-coated TiO 2 NPs (100 ppm).Fluorescence spectra were acquired, and oxidation was quantified by monitoring the spectral shift from red ( max = 594 nm) to green ( max = 520 nm).These experiments were also performed in the presence of ROS scavengers, i.e., D-mannitol ( • OH scavenger) [39] and SOD ( • O 2 − scavenger), [40] to obtain information on the role of different ROS on lipid oxidation.All measurements were performed in duplicate at 37 °C.

Quartz Crystal Microbalance with Dissipation Monitoring (QCM-d)
QCM-d measurements were performed as described in detail before. [22]In short, these measurements were performed using a QSense Analyzer equipped with both standard and UVtransparent window modules (Biolin Scientific, Sweden).Supported bilayers were prepared by SUVs' deposition on silicon dioxide surfaces (QSense QSX 303 SiO 2 , 4.95 ± 0.05 MHz).For this, sample cells and tubing were cleaned with 2% Hellmanex and multiple MQ rinses under bath sonication, followed by rinsing with ethanol and N 2 drying.After similar cleaning, the SiO 2 surfaces were plasma cleaned (PDC-32G, Harrick Plasma, USA) for 2 min before cell assembly.SUVs (0.1 mg mL −1 in MQ) were injected into the chamber and their deposition, rupture, and full bilayer formation were monitored from the variation of frequency shift and dissipation changes (ΔF and ΔD, respectively).
After rinsing with MQ water to remove non-adsorbed vesicles, the chamber was flushed with a buffer solution containing either bare or peptide-coated TiO 2 NPs (100 ppm).Subsequently, samples were exposed to UV light at a distance of 6 cm (Spectroline ENF-260C, 254 nm; 3 mW cm −2 ) to trigger oxidation.All measurements were performed at 37 °C at least in triplicate.

Neutron Reflectometry (NR)
NR experiments were performed as described in detail previously. [22]In brief, experiments on +PG bilayers interacting with bare and LL-37-coated TiO 2 NPs were performed on the D17 vertical reflectometer (Institute Laue-Langevin, Grenoble, France), operating in time-of-flight mode. [41,42]The Q-region of interest (0.01 to 0.3 Å −1 ) was accessed using two incident angles (0.8°and 3.0°), wavelength range between 2 and 30 Å and wavelength resolution between 1 and 4%.For kinetic measurements, an intermediate angle of 1.8°was used to monitor the Q-region where the main changes in the reflectivity profile were expected to occur, and a divergent beam geometry was set to achieve improved statistics with shorter acquisition times (60 s).Experiments on PC bilayers interacting with LL-37-coated TiO 2 NPs were performed using the OffSpec reflectometer (ISIS Pulsed Neutron and Muon Source, Rutherford Appleton Laboratory, Harwell, UK). [43,44]Three incident angles (0.3°, 1.0°a nd 2.3°) were used to cover the Q-region from ≈0.01 to 0.35 Å −1 .For kinetic measurements, the reflectometry changes at the first two angles were recorded every 10 min, acquiring data for 30 s at 0.3°and for 9 min 30 s at 1.0°.Solid-liquid flow cells, the top plate of which was modified with a 30 mm diameter circular opening to allow in situ UV irradiation were used, together with UV-transparent quartz blocks (80×50×15 mm, 1 face polished, RMS < 4. The bare quartz surfaces were first characterized in D 2 O and MQ.Then, tip-sonicated SUVs (0.1 mg mL −1 ) were injected by syringe and allowed to deposit for 20 min, after which the excess was rinsed off with 10 mL MQ and 10 mL of the desired buffer at 2 mL min −1 .The bilayers thus formed were characterized in two contrasts, 10 mM acetate in MQ (HAc) or in D 2 O (DAc), before treatment.Contrast exchange was performed by pumping 20 mL of the desired buffer at 2 mL min −1 , all buffers being degassed by bath sonication prior to experiments.Samples were incubated for 10 min with bare or LL-37-coated TiO 2 NPs (100 ppm) in DAc at the desired pH.The bilayers were then subjected to in situ UV irradiation (Spectroline ENF-260C, 254 nm; 3 mW cm −2 ) for 2 h.During this time, kinetic measurements were taken every minute in the middle-Q region as mentioned above.Immediately after UV exposure, the whole Q-range was measured in DAc, after which the bilayer was rinsed and characterized in both contrasts.Experimental NR profiles were fitted by using the Genetic Optimization method available on the Motofit software within the analysis package IGOR Pro. [45,46]A series of parallel layers were used to model the interfacial structure, each of these described by a set of physical parameters, including thickness, roughness, hydration, and neutron scattering length density (SLD).The best fits of these parameters were then converted into SLD profiles, which represent the density distribution in the direction perpendicular to the reflecting interface.A Monte Carlo error analysis allowing for refitting data 200 times was employed to minimize the uncertainty associated with data fitting. [47]9.Live/Dead Bacterial Viability Assay E. coli ATCC 25922 and S. aureus ATCC 29213 were stained using the LIVE/DEAD BacLight Bacterial Viability Kit (Thermo Fisher Scientific Inc., Waltham, USA).Bacteria were grown to stationary phase in 25 mL Lennox broth (LB Broth; Sigma Aldrich (St.Luis, USA)) overnight at room temperature, shaking at 180 rpm.The bacteria were pelleted and washed by centrifugation (10 000 × g, 10 min, twice) and re-suspended in 10 mM acetate buffer, pH 5.4.Nanoparticle dispersions, either bare or loaded with LL-37 at their maximum loading capacity, were added to bacteria for a final TiO 2 NPs concentration of 100 ppm (10 μM LL-37) and an optical density OD 600 of 1.2, corresponding to 12 × 10 8 colony forming units (CFU) mL −1 for E. coli and 6 × 10 8 (CFU) mL −1 for S. aureus (confirmed on subsequent dilution, plating, culture, and counting of colonies), in the final mixture.Samples were then placed in quartz cuvettes and incubated for 1 hour at room temperature, either in darkness or in the presence of UV illumination (Spectroline ENF-260C, 254 nm; 3 mW/cm 2 ), the latter at a cuvette-lamp distance of 6 cm.Samples were subsequently diluted in 10 mM acetate buffer, pH 5.4, to obtain an OD 600 of 0.6, corresponding to 6 × 10 8 (CFU) mL −1 for E. coli and 3 × 10 8 (CFU) mL −1 for S. aureus.This was followed by 10 min of staining of 200 μL of sample with 0.5 μL of a 1/1 (v/v) mixture of the fluorescent probes SYTO 9 (excitation/emission maxima 480/500 nm) and propidium iodide (excitation/emission maxima 490/635 nm).[48] Subsequently, samples were imaged through confocal microscopy.

Confocal Microscopy
Bacteria were plated onto a cover slide at 10 8 CFU mL −1 and then imaged with a 100×/1.25 oil objective using a Leica DMi8 confocal microscope (Leica Microsystems, Washington, D.C., USA).For each sample, 20 randomized, wide-field images (100 μm × 100 μm) were collected.Quantification of the fraction of Live/Dead bacteria was performed with the software ImageJ (National Institutes of Health, Bethesda, USA). [36,37]Experiments were run in triplicate at 25 °C.

Lactate Dehydrogenase (LDH) Assay
To investigate cell toxicity, THP1 monocytes (THP1-Xblue-CD14 reporter cells (InvivoGen, France) were cultured according to the manufacturer's instructions, and nanoparticle-induced release of the intracellular enzyme LDH was monitored.For this, Hundred ppm bare or LL-37-coated TiO 2 NPs were mixed with 100 ppm (0.1 mg mL −1 ) lipopolysaccharide (LPS) from E. coli O111:B4 in 10 mM Tris, pH 7.4.Samples were then placed in quartz cuvettes and incubated for 2 hours at room temperature, either in darkness or in the presence of UV illumination (Spectroline ENF-260C, 254 nm; 3 mW/cm 2 ), at a cuvette-lamp distance of 6 cm.Next, cells (1 × 10 6 cells mL −1 ) were incubated with LPS/bare TiO 2 or LPS/LL-37-coated TiO 2 .To be able to compare results for cells and bacteria, LDH experiments were performed at an average of 250 TiO 2 NPs per cell, which is comparable to the range used for the confocal microscopy studies of bacteria, the latter being in the range of 200 (E.coli) to 400 (S.aureus) nanoparticles per CFU.After 22-24 h incubation at 37 °C in 5% CO 2 , LDH release was measured in triplicate using a lactate dehydrogenase assay kit (Invitrogen CyQUANT LDH Cytotoxicity Assay, Fisher Scientific), according to manufacturer's instructions.

Statistical Analysis
All data were presented as means with standard errors.For each experiment (except for neutron reflectometry experiments), measurements were performed at least in triplicate.Errors associated with the bilayer's structural parameters from neutron reflectometry were obtained through a Monte Carlo error analysis, [47] available on the Motofit software within the analysis package IGOR Pro, [45,46] which allowed for refitting experimental NR profiles 200 times, to minimize the uncertainty associated to data fitting.

Peptide Coating of TiO 2 NPs
TiO 2 NPs have previously been reported to display pH-dependent charge density, with an isoelectric point at pH 6-6.5. [49,50]In line with this, TiO 2 NPs were found to display a net positive -potential at pH 3.4 and 5.4, and a negative charge at pH 7.4 and 9.4 (Figure 1A). [23]Mirroring this, TiO 2 nanoparticles displayed pronounced aggregation close to the isoelectric point, while much less so at pH 3.4 and 9.4.By loading TiO 2 NPs with LL-37 with increasing peptide concentrations at pH 5.4, an increasingly positive  -potential was measured, suggesting a concentration-dependent peptide binding to the TiO 2 nanoparticle surface (Figure 1B).For TiO 2 NPs fully loaded with LL-37 (≥10 μM), the -potential reaches +38 ± 2 mV, providing improved colloidal stability to the peptide-loaded nanoparticles.Furthermore, the LL-37 coating was found to be relatively resilient to UV degradation, as indicated by only partially suppressed positive  -potential after up to 2 h of UV illumination (Figure 1C).At pH 7.4, similar effects were observed, although a higher peptide concentration was needed to achieve saturation in positive potential and colloidal stability (Figure S2A,B, Supporting Information).In contrast to partial peptide layer degradation on prolonged UV illumination for peptide-coated TiO 2 NPs, no such effects were observed for non-photocatalytic mesoporous SiO 2 nanoparticles under these conditions (Figure S3A,B, Supporting Information), demonstrating that UV itself (i.e., in the absence of TiO 2 NPs) does not induce any significant peptide degradation or oxidation under the experimental conditions of the study.

ROS Generation and Lipid Oxidation
Since the adsorbed peptide layer may detrimentally influence the efficiency by which photogenerated electrons and holes reach the particle/water interface to be able to react with water and oxygen, key for ROS generation, we next investigated how peptide coating affected lipid oxidation in PC, +Chol and +PG large unilamellar vesicles (LUVs) by TiO 2 NPs.As shown in Figure S4 (Supporting Information), oxidation was very low (3-5% min −1 ) in the absence of TiO 2 NPs.When bare TiO 2 NPs were present, the oxidation rate was 4-6 times higher for +PG than that for PC or +Chol, the latter showing similar oxidation rates as in the absence of nanoparticles at pH 5.4 (Figure 2; see also Figure S5 (Supporting Information) for corresponding oxidation kinetics) and 7.4 (Figures S7 and S8A, Supporting Information, for oxidation rates and oxidation kinetics, respectively).Importantly, LL-37 coating did not markedly influence ROS generation, indicating that the peptide layer was sufficiently "open" to allow photogenerated electrons and holes to reach the particle surface to form ROS. Furthermore, the addition of D-mannitol ( • OH scavenger) [39] to LUV suspensions resulted in a significant suppression of the oxidation  − inhibitor) [40] had a limited impact on the oxidation of bare TiO 2 NPs (Figure 2  While C 11 -BODIPY experiments provide valuable information on UV-induced ROS generation and subsequent oxidation triggered by these, these report on oxidation of the dye, incorporated into lipid bilayers, rather than on oxidation of the bilayer phospholipids.Considering this, QCM-d was also included in our investigation to directly monitor the consequences of lipid oxidation.For this, lipid bilayers from PC, +Chol, or +PG SUVs were formed on SiO 2 -coated quartz crystals.The successful formation of homogeneous lipid bilayers was confirmed by changes in frequency (ΔF ≈ −25-28 Hz) and dissipation (ΔD ≈ 0) (Figure S10, Supporting Information). [51]LL-37-coated TiO 2 NPs were found to bind to all the membranes investigated (Figure 3A and S11A for representative QCM-d profiles).Quantitatively, however, nanoparticle deposition was much higher on anionic +PG bilayers + (−66 ± 13 Hz), than on zwitterionic PC or +Chol membranes (−32 ± 4 and −6 ± 2 Hz, respectively).In contrast, bare TiO 2 nanoparticles did not bind to +PG membranes (+ 0.5 ± 2 Hz, Figure 3B).A discussion of the binding of bare TiO 2 NPs to PC, +PG, and +Chol membranes, as well as photocatalytic effects under UV illumination can be found in. [22]Thus, the extensive adsorption of LL-37-coated TiO 2 NPs to anionic +PG membranes can be directly attributed to the cationic peptide coating of the NPs.A possible adsorption of free peptide (i.e., not bound to the NPs surface), leading to comparable shifts in the frequency, was ruled out through a control experiment (Figure S12, Supporting Information).
After NP binding, the bilayers were exposed to UV illumination (Figure 4A,B).Mirroring the result on NP binding, only minor variations of frequency (i.e., |ΔF|<7 Hz) and dissipation (i.e., |ΔD|<1•10 −6 ) were found for PC and +Chol bilayers after UV exposure for 2 h in the presence of LL-37-loaded TiO 2 NPs  (Figure 4A; see also representative QCM-d profiles in Figure S11B, Supporting Information).In contrast, a dramatic increase in frequency (38 ± 5 Hz) was observed for +PG, indicating a massive loss of material from the surface.In parallel, significant decreases in dissipation were observed, suggesting the formation of a more diffuse adsorbed layer (Figure S11B, Supporting Information).These results indicate that LL-37-coated TiO 2 NPs induce major structural rearrangements of +PG bilayers, resulting from UV-induced oxidative degradation, but very limited structural destabilization of PC and +Chol membranes.Similar results were obtained at pH 7.4, showing that LL-37 coatings enhanced nanoparticles binding to +PG (Figure S13A, Supporting Information), as well as UV-induced oxidative degradation (Figure S13B, Supporting Information).

Consequences of Lipid Oxidation for Membrane Structure and Destabilization
To obtain more detailed clues to how nanoparticle-induced oxidation translates into structural changes in the lipid bilayers, NR experiments were performed.For this, experiments were carried out at pH 5.4, using acetate buffer prepared either in H 2 O (HAc) or D 2 O (DAc).Results obtained under the two contrasts were fitted simultaneously to a 4-layer model (Model 1 in Scheme S1), with the input parameters reported in Tables S1 and S2 (Supporting Information).To reduce the number of free parameters, the head group thickness was fixed at 7.5 Å. [52] For all three membrane compositions, the initial bilayers presented >99% coverages and low (≈0%) tail hydration.The best-fitting parameters for the +PG and PC bilayers are reported in Table S2 (Supporting Information).Values of area per molecule (APM) and surface coverage (Γ), calculated as described previously, [22] are consistent with previous literature. [53,54]hown in Figure 5 are reflectivity data with best curve fits and corresponding SLD profiles for +PG bilayers interacting with 100 ppm bare TiO 2 (left panel) or LL-37-TiO 2 NPs (middle panel), and for PC bilayers interacting with LL-37-TiO 2 NPs (right panel), before and right after 2 hours of UV illumination.Reflectivity data and corresponding SLD profiles of +PG and PC bilayers before NP addition are also shown for comparison.Key structural data obtained from neutron reflectometry fits are plotted in Figure 6.Structural effects on supported +PG bilayers with bare TiO 2 or LL-37-TiO 2 NPs, and on supported PC bilayers with LL-37-TiO 2 NPs.The reported parameters were extracted from neutron reflectometry fits for the three systems: [1] before nanoparticle incubation; [2] after nanoparticle incubation; [3] S2 (Supporting Information).Without UV illumination, no structural modifications were detected for +PG bilayer exposed to bare TiO 2 NPs (Figure 5, left panel), as evidenced by the preservation of its initial structural parameters (Figure 6).In contrast, the incubation with LL-37-TiO 2 NPs (prior UV illumination) caused significant modifications of the +PG structure (Figure 5, middle panel).Specifically, the area per molecule (APM) for +PG changed from 64 ± 3 Å 2 before nanoparticle addition to 173 ± 7 Å 2 after the addition of TiO 2 NPs coated with LL-37, which was accompanied by increased hydration of both the polar headgroups and acyl chain regions of the bilayer (Figure 6).A considerable lipid removal was associated with this, with surface coverage decreasing from 4.0 ± 0.1 mg m −2 for the initial bilayer, to 1.5 ± 0.1 mg m −2 after the incubation with LL-37-TiO 2 NPs.In addition, a rough NP outer layer was formed on the surface of the lipid bilayer, as reported also previously [22] and sketched in Scheme S1 (Model 2), the latter de-scribed by the structural parameters listed in Table S3 and plotted in Figure S14 (Supporting Information).In line with QCM-d results (Figure 3B), the occurrence of this layer (absent for +PG interacting with bare TiO 2 NPs) is connected to the much higher adsorption of LL-37-coated NPs as compared to bare ones.In contrast to the large effects observed for +PG bilayers, only minor variations in the reflectivity profiles and corresponding SLDs were observed for PC (Figure 5, right panel), associated with negligible changes in the AMP and surface coverage (Figure 6, and Table S2, Supporting Information).The NPs outer layer, formed upon LL-37-TiO 2 NPs adsorption, showed slightly lower thickness and higher hydration as compared to the one formed on +PG bilayers (Figure S14, and Table S3, Supporting Information), highlighting lower adsorption of LL-37-TiO 2 NPs, in line with QCM-d data (Figure 3A).
For +PG bilayers, some structural variations were observed as induced also by bare TiO 2 NPs upon 2 hours of UV illumination, with an increase of the AMP from 64 ± 3 Å 2 to 89 ± 4 Å 2 (Figure 6).This corresponds to a partial bilayer removal, with the final surface coverage changing from 3.98 ± 0.04 mg m −2 (before UV) to 2.8 ± 0.1 mg m −2 (after UV).However, a much more pronounced destabilization of +PG was observed upon UV illumination in the presence of LL-37-TiO 2 NPs, with the hydration of the acyl chains increasing from 48 ± 4% to 88 ± 1%, paralleled by a similar increase in the polar headgroups hydration (from 74 ± 10% to 87 ± 2%).As a consequence of this, the final APM values increased to 419 ± 7 Å 2 , corresponding to an almost quantitative bilayer removal, with a final surface coverage of 0.60 ± 0.06 mg m −2 .In contrast, LL-37-TiO 2 NPs only caused a negligible UV-induced structural destabilization of PC bilayers, leaving all key structural parameters of the lipid bilayer essentially unaffected.This demonstrates that LL-37-TiO 2 NPs show a high selectivity for +PG bilayers compared to PC in terms of UV-induced oxidative degradation, in line with the QCM-d results on the same systems (Figure 4A).
Next, reflectivity changes were monitored during UV illumination over a narrower Q-range for faster data acquisition. [55]Results for +PG interacting with bare or LL-37-TiO 2 NPs are shown in Figure S15 (Supporting Information) while corresponding data for PC are shown in Figure S16 (Supporting Information).As shown in Figure S15 (Supporting Information) (top panel), relatively small effects were observed for bare TiO 2 NPs.In contrast, pronounced decreases in reflectivity and correspond-ing SLD variations were observed over time for LL-37-TiO 2 NPs (Figure S15, Supporting Information, bottom panel).Analysis of these results (Figure 7) showed that the destabilization of +PG induced by LL-37-TiO 2 NPs under UV illumination involved an initial incubation phase of ≈ 20 min, characterized by relatively stable values of headgroups hydration, AMP, and surface coverage, paralleled by mild variations in the acyl chains hydration.After that, a fast oxidative lipid removal, mirrored by a steep increase in the AMP, was observed on subsequent UV exposure.This process slowed down only after 100 min of UV exposure when the surface coverage value approached 0 mg m −2 .Remarkably, the acyl chains region underwent significant changes in the first 40 min of UV exposure as compared to the polar headgroups, suggesting that the oxidative destabilization induced by LL-37-TiO 2 NPs upon UV originated in the hydrophobic bilayer core.A similar analysis for PC bilayers interacting with LL-37-TiO 2 NPs is reported in Figure S17 (Supporting Information), showing no detectable variations in the bilayer structural parameters throughout 2 hours of UV exposure.

Antibacterial Effects and Cell Toxicity
In order to investigate if the results for model lipid bilayers translated to bacterial and human cell membranes, we next investigated antimicrobial effects of bare and peptide-coated TiO 2 NPs, as well as their toxicity against human cells.In order to monitor antimicrobial effects, confocal microscopy was employed to study the effects of bare and LL-37-coated TiO 2 NPs on E. coli and S. aureus bacteria.For this, a two-color fluorescence LIVE/DEAD assay was used, which allows for separating the fluorescence from bacteria with intact cell membranes (staining green) from the ones with damaged membranes (staining red).Representative confocal microscopy images, separately showing intact (greenstained) and dead (red-stained) bacteria, and corresponding data, are shown in Figures 8 and 9A, respectively.In the absence of UV illumination, bare TiO 2 NPs had negligible effects on Gramnegative E. coli (3 ± 1% dead bacteria), while displaying some antimicrobial action against Gram-positive S. aureus (24 ± 6% dead bacteria).For both E. coli and S. aureus, the decrease in viability observed upon UV illumination was similar in the presence and absence of bare TiO 2 NPs, indicating relatively minor effects of the latter.In contrast, LL-37-coated TiO 2 NPs strongly promoted bacteria killing already prior to UV illumination, particularly so for E. coli, leading to a dead bacteria percentage of 68 ± 5% for E. coli and 54 ± 1% for S. aureus.UV illumination further enhanced the antimicrobial efficiency of LL-37-TiO 2 NPs, leading to 88 ± 5% and 69 ± 3% dead E. coli and S. aureus, respectively.
Possible cytotoxic effects of nanoparticles against eukaryotic cells were addressed by lactate dehydrogenase experiments em-ploying human THP1 monocytes.For this, THP1 monocytes were exposed to bare and LL-37-coated TiO 2 NPs with or without UV illumination.Contrasting the boosting of antimicrobial effects by LL-37 coating, lactate dehydrogenase experiments demonstrated that both bare and LL-37-coated TiO 2 NPs (either before or after UV illumination) did not induce any significant increase in membrane rupture and resultant release of intracellular lactate dehydrogenase (Figure 9B).Together, these results indicate that the boosted photocatalytic destabilization by LL-37-coated TiO 2 NPs, found for bacteria-mimicking model lipid membranes, translate also to the biological systems.Furthermore, they demonstrate that AMP coating indeed provides an approach to increase the selectivity of photocatalytic TiO 2 NPs to bacterial membranes.

Discussion
Cationic surface modification of nanomaterials is a frequently employed approach for enhancing their antimicrobial effects.[5] At the same time as cationization may enhance antimicrobial effects under some conditions, however, unselective binding of anionic proteins may cause corona formation, [56,57] which in turn may result in decreased antimicrobial effects.In addition, cationic surface modification frequently results in destabilization also of human cells, thereby inducing toxicity. [3,58,59]To increase selectivity of such surface modification of antimicrobial nanomaterials, it is therefore key to search for those that are able to selectively boost nanoparticle destabilization of bacterial membranes, leaving membranes of human cells unaffected.Of interest to the present study, AMPs may display potent antimicrobial effects, also against strains resistant to antibiotics. [25,28,29]It is therefore interesting to investigate to what extent such selectivity can be translated to antimicrobial nanomaterials by their surface coating by such peptides.Here, it should be remembered, however, that antimicrobial effects of AMPs rely on peptide binding and insertion into the membrane, sometimes associated with conformational changes, as well as effects on anionic lipid recruitment, domain formation, and flip-flop rates. [25]As these effects can all be expected to be partially impaired by constraining the AMP to the particle surface, it is unclear how much of the selectivity displayed by the free peptide can be transferred to the coated nanoparticles.Considering this, different approaches have been reported for increasing antimicrobial effects of surface-bound AMPs, notably by tethering/loading the peptides to surface-bound polymer chains/matrices, thereby increasing the mobility freedom of the AMP, approaching that of the free peptide. [60,61]While little is known on how peptide orientation, packing density, mobility freedom, and conformational changes for surface-bound AMPs affect their antimicrobial effects, and the selectivity between bacteria and human cells, AMP coating of nanomaterials has been reported in a couple of studies to increase antimicrobial potency and selectivity, [62,63] as well as to provide additional features of interest, such as aggregation of bacteria and bacterial lipopolysaccharides for localization of infection and inflammation. [64]Similarly, anti-cancer effects displayed by some AMPs in solution may be transferred to nanoparticles by peptide coating for selective targeting of cancer cells. [65]or photocatalytic nanomaterials, boosting antimicrobial effects and membrane selectivity by AMP coating is, however, more complex than for other nanomaterials.Thus, the peptide coating may result in an effectively reduced particle/water surface, and hence also reduced possibilities for ROS formation.Furthermore, the peptide may be degraded by generated ROS, and the particle coatings are effectively destroyed on illumination.Despite these concerns, a couple of studies have reported advantageous effects on coating photocatalytic nanomaterials with antimicrobial peptides and other macromolecules.Thus, Lu et al. reported dimethylpyrrolidonium-coated C 60 to strongly suppress bacteria in wounds infected with P. aeruginosa during illumination.This, in turn, resulted in a dramatic improvement in survival, from 8% to 82% survival after treatment. [66]imilarly, Zhang et al. showed peptide-fullerene hydrogels to potently suppress multi-drug resistant S. aureus and promote wound healing. [67]Moreover, Chen et al. found ZnO nanoparticles coated by the AMP UBI 29-41 to enhance the antimicrobial effect of vancomycin against S. aureus and B. subtilis. [68]Extending on these studies, as well as on our previous studies on bare TiO 2 nanoparticles, [22,23] the present investigation provides a mechanistic foundation for observations of retained or improved antimicrobial effects of photocatalytic nanoparticles after AMP coating.In particular, it shows that peptide coating may be done without substantially suppressing or altering ROS generation on illumination.While the  -potential results indicate that the peptide layer on TiO 2 NPs undergoes some oxidation on UV exposure, which may include surface-bound peptide participating in the reaction through the generation of reactive oxygen or nitrogen species, particle size and -potential results also show peptidecoated dispersions to be colloidally stable, and the layers thus stable with respect to peptide desorption over the duration of the experiments (≈hours).Hence, there is no experimental support for the peptide layers effectively desorbing to catalyze oxidative degradation of either model lipid or bacteria/cell membranes.Instead, the QCM-d results clearly show that the main effect of the peptide coating (remaining positively charged also after UV exposure) is to promote binding to bilayers enriched in anionic PG.Due to the short lifetime of ROS, close proximity between ROS generated and their lipid target molecules affords AMP-coated TiO 2 nanoparticles with promoted capacity to destabilize lipid membranes under UV illumination.Apart from outlining the details of such oxidative destabilization of lipid membranes in detail for model lipid systems, the antimicrobial and cell toxicity results provided in this investigation demonstrate such effects to translate also to the biological situation.
However, there are still numerous issues that will have to be resolved regarding the selectivity of peptide-coated photocatalytic nanoparticles.For example, qualitatively similar effects were observed for the AMP LL-37 and for poly(arginine) homopolypeptide (Figure S18, Supporting Information), but it remains to be investigated how the different antimicrobial potency and selectivity displayed by these as free peptides in solution [69][70][71] translate into more detailed biological performance.Furthermore, the antimicrobial action of LL-37 has been correlated to the formation of an amphiphilic helix conformation, [69] an effect likely to be (partly) lost after peptide immobilization at the nanoparticle surface.Considering this, membrane interactions may be governed by the distribution of charges, as well as hydrophilic and hydrophobic amino acids, at the surface of peptide-coated nanoparticles.Addressing these aspects is therefore a key issue in further studies.
In addition, bacterial lipopolysaccharides play a key role not only in the structural integrity of bacteria but also in their pathogenicity.Investigating the degradation of lipopolysaccharides by bare as well as peptide-coated nanoparticles is therefore key for a complete understanding of photocatalytic antimicrobial effects of such systems.So far, only a couple of studies have been reported, suggesting that lipopolysaccharides provide further complexity to photocatalytic antimicrobial effects of nanoparticles by: i) having a lower susceptibility to photocatalytic degradation than phospholipids, and ii) providing a barrier affecting the proximity between nanoparticles and the plasma membrane of bacteria. [9]In this context, AMPs again offer interesting opportunities since some of these have been found to be able to bind bacterial lipopolysaccharides and suppress their inflammatory properties, [26,72,73] a key feature for suppressing inflammation following the generation of bacterial debris after membrane lysis by antimicrobial peptides.
From a wider perspective, we note that the approach taken in the present study to boost the antimicrobial effect by promoting proximity between bacteria and antimicrobial nanoparticles may also have a bearing also for other pathogens, other types of nanoparticles, and even other biological questions.For example, photocatalytic nanoparticles have been found to display potent antiviral effects on illumination, [8,9,74,75] hence surface modifying these with compounds having an affinity for virus capsids or capsid envelopes, such as aptamers, antiviral peptides, or antibodies toward capsid proteins may offer a way to boost antiviral effects of photocatalytic nanoparticles without simultaneously increasing cell toxicity.Indeed, this is supported by findings by Hu et al., who immobilized aptamers to GO and found this to promote virus binding and result in boosted inactivation of bacteriophage MS2 via oxidative damage under illumination. [76]Analogously, one could imagine that bacteria targeting can promote antimicrobial effects also for non-photocatalytic nanoparticles. [3]xemplifying this for nanoparticles causing membrane disruption by particle binding, Malekkhaiat Häffner et al. coated spiky, virus-like, mesoporous silica nanoparticles with the antimicrobial peptide LL-37 and found peptide coating to strongly boost membrane disruption and antimicrobial effects caused by such spiky nanoparticles, [35] analogous to edge effects caused by sheet nanoparticles. [77]Similarly, one could envision nanoparticles reaching antimicrobial effects through local heating induced by either light or magnetic field [3] to benefit from enhanced proximity between bacteria and nanoparticles through targeting surface coatings, e.g., based on antimicrobial peptides. [78,63]Finally, we note that photocatalytic and other nanomaterials are interesting not only for combatting infection but also in other biological contexts, such as in tumor therapeutics.Exemplifying this, Duong et al. demonstrated doxorubicin-containing polymer nanoparticles coated with the antimicrobial peptide GRR10W4 to display improved uptake and toxic effects for melanoma cancer cells, but not for non-malignant fibroblasts and keratinocytes. [65]Considering this, as well as the promising results in the present study, investigations into these lines of inquiry seem relevant.
From a mechanistic perspective, key developments in nanoparticle-membrane interactions for other types of nanoparticles may be of relevance also for photocatalytic NPs, either bare or coated. [9,79,80][81][82] Furthermore, preferential nanoparticle interactions with membrane domains [83,84] are likely important for photocatalytic NPs and deserve further attention, as do the effects of nanoparticles on integrin activation and other key membrane processes by systematically varied nanoparticle properties. [85]From this perspective, 4, the peptide coating does not detrimentally interfere with ROS generation and displays good stability on UV exposure for 1-2 h.As a result of this, binding of peptide-coated TiO 2 is much higher than that of bare TiO 2 NPs, particularly at anionic bacteria-like membranes, and oxidative degradation of the latter on UV exposure strongly boosted by peptide coating, whereas corresponding effects on zwitterionic mammalian-like membranes were much smaller.Mirroring this, LL-37-coated TiO 2 NPs displayed boosted antimicrobial effects against both Gram-negative E. coli and Gram-positive S. aureus bacteria, whereas toxicity against human THP1 monocytes remained low.
we note that the platform described in the present investigation is compatible with all such future research directions.

Conclusions
As indicated by -potential measurements, TiO 2 -bound LL-37 remains partially intact even after UV illumination for 1-2 hours.Furthermore, ROS generation on illumination of TiO 2 NPs is not detrimentally affected by AMP coating, as assessed from C 11 -BODIPY fluorescence.QCM-d showed that coating with positively charged peptides promotes the binding of TiO 2 NPs to anionic bacteria-mimicking membranes, thereby boosting oxidative membrane degradation during UV illumination.In contrast, zwitterionic mammalian-like membranes displayed much lower binding of LL-37-coated TiO 2 NPs and were much less susceptible to oxidative degradation.NR results showed such membrane destabilization to involve increased hydration, lipid removal, and solubilization.Mirroring these physicochemical results obtained for model lipid membranes, antimicrobial effects under UV illumination were found to be promoted after LL-37 coating for both E. coli and S. aureus.Meanwhile, toxicity against human THP1 monocytes remained low for the LL-37-coated TiO 2 NPs.Together, these findings illustrate that AMP coating offers opportunities for selectively boosting the antimicrobial effects of photocatalytic nanoparticles at retained low toxicity (Figure 10).

Figure 1 .
Figure 1.A)  -potential and number average particle size of TiO 2 NPs as a function of pH for 100 ppm TiO 2 in 10 mM acetate pH 3.4 and pH 5.4 or 10 mM Tris pH 7.4 and 9.4.B) Corresponding results for 100 ppm TiO 2 NPs loaded at varying concentrations of LL-37 in 10 mM acetate, pH 5.4.C)  -potential of TiO 2 NPs (100 ppm) coated with LL-37 (10 μM) in 10 mM acetate, pH 5.4, before and after UV illumination for 1 and 2 h.Measurements were performed in triplicate.
5 Å, PI-KEM Ltd., Tamworth, UK).The blocks were cleaned in dilute acid piranha solution (5/4/1 v/v H 2 O/H 2 SO 4 /H 2 O 2 ) at 80 °C for 15 min, rinsed thoroughly in MQ, followed by 10 min of UV-ozone treatment (UV/Ozone ProCleaner, BioForce Nanosciences, USA).HPLC tubing, PEEK troughs, and O-rings were cleaned with a 2% Hellmanex solution (Hellma Analytics, UK) in bath sonication, then both thoroughly rinsed and sonicated in MQ.A circulating water bath was used during the measurements to keep the sample cells at 37 °C.
) and 7.4 (Figures S7 and S8B, Supporting Information), showing • OH radicals to be formed during UV illumination.In contrast, SOD (•O 2 and Figure S6B, Supporting Information, for pH 5.4, and Figures S7 and S8C, Supporting Information for pH 7.4).In line with this, experiments under de-aerated conditions gave results similar to those obtained under aerated conditions (Figure S9A,B, Supporting Information), showing the effects of dissolved oxygen to be relatively minor.Analogous effects of Dmannitol and SOD were observed for LL-37-TiO 2 NPs at both pH 5.4 (Figure 2 and Figure S6A,B, Supporting Information) and pH 7.4 (Figures S7 and S8B,C, Supporting Information).

Figure 4 .
Figure 4. A) QCM-d results showing the Frequency Shift (Hz) induced by LL-37-TiO 2 NPs on supported PC, +Chol, and +PG bilayers with or without UV illumination.B) Corresponding effects of bare and LL-37-TiO 2 NPs on supported +PG bilayers.All measurements were performed in 10 mM acetate, pH 5.4, in triplicate.Representative QCM-d profiles are reported in Figure S11B (Supporting Information).

Figure 5 .
Figure 5. Neutron reflectivity curves with best model fit (upper) and corresponding SLD profiles (lower) for supported +PG bilayers before and after incubation with bare TiO 2 NPs (left), LL-37-coated TiO 2 NPs (middle), or for supported PC bilayers before and after incubation with LL-37-coated TiO 2 NPs (right).Shown are also the reflectivity curves with the best model fit and SLD profiles for the corresponding systems after 2 h of in situ UV exposure.All experiments were performed in 10 mM acetate buffer, pH 5.4, at a nanoparticle concentration of 100 ppm.Curves are shown for two different buffer contrasts (DAc and HAc) and the data for the latter are offset by 1•10 −5 for clarity.A small offset is also applied between NR profiles in the same buffer, obtained before, after NPs addition, and after 2 h of UV illumination.The grey box in the SLD profiles indicates the position of the silicon block and reflecting interface, consisting of bulk Si and a SiO 2 layer.
Figure6.Structural effects on supported +PG bilayers with bare TiO 2 or LL-37-TiO 2 NPs, and on supported PC bilayers with LL-37-TiO 2 NPs.The reported parameters were extracted from neutron reflectometry fits for the three systems:[1] before nanoparticle incubation;[2] after nanoparticle incubation;[3] after 2 h in situ UV exposure.Shown are changes in A) bilayer thickness, B) area per lipid molecule (APM), C) surface coverage of the supported lipid bilayer (Г), D) hydration of the hydrophilic headgroups, and E) hydration of the hydrophobic tails.Experiments were performed in 10 mM acetate buffer, pH 5.4, at a nanoparticle concentration of 100 ppm.

Figure 6
Figure 6 and listed in full in TableS2(Supporting Information).Without UV illumination, no structural modifications were detected for +PG bilayer exposed to bare TiO 2 NPs (Figure5, left panel), as evidenced by the preservation of its initial structural parameters (Figure6).In contrast, the incubation with LL-37-TiO 2 NPs (prior UV illumination) caused significant modifications of the +PG structure (Figure5, middle panel).Specifically, the area per molecule (APM) for +PG changed from 64 ± 3 Å 2 before nanoparticle addition to 173 ± 7 Å 2 after the addition of TiO 2 NPs coated with LL-37, which was accompanied by increased hydration of both the polar headgroups and acyl chain regions of the bilayer (Figure6).A considerable lipid removal was associated with this, with surface coverage decreasing from 4.0 ± 0.1 mg m −2 for the initial bilayer, to 1.5 ± 0.1 mg m −2 after the incubation with LL-37-TiO 2 NPs.In addition, a rough NP outer layer was formed on the surface of the lipid bilayer, as reported also previously[22] and sketched in Scheme S1 (Model 2), the latter de-

Figure 7 .
Figure 7. Structural parameters, obtained from neutron reflectometry fits (Figure S15, Supporting Information), of +PG supported lipid bilayer before, and after the incubation of either bare or LL-37-coated TiO 2 NPs, as well as at different time points during exposure to UV illumination.Shown are the results of the calculated A) adsorbed amount (Г), B) the area per molecule (APM), C) as well as of the hydration of the hydrophilic headgroups and D) hydrophobic tails.

Figure 8 .
Figure 8. Representative confocal microscopy images obtained using LIVE/DEAD assay (red, green, and Differential Interference Contrast images overlaid) for 10 8 CFU mL −1 of E. coli and S. aureus bacteria in 10 mM acetate, pH 5.4, without or with 1 hour of incubation with bare (middle) or LL-37-coated (right) TiO 2 NPs with or without UV illumination.For each system, the fluorescence from alive (green) and dead (red) bacteria is shown in separate images (left images for alive bacteria and right images for dead ones).

Figure 9 .
Figure 9. A) Quantification of confocal microscopy images, showing percentages of dead E. coli and S. aureus bacteria with or without UV illumination.Results are shown both for untreated bacteria and for bacteria exposure for 1 hour to TiO 2 or LL-37-TiO 2 NPs.B) LDH release from human THP1 monocytes in the absence and presence of bare TiO 2 or LL-37-coated TiO 2 NPs, with or without UV illumination.The dashed line represents the background release in buffer.Measurements were performed in triplicate.

Figure 10 .
Figure 10.Schematic illustration of experimental results demonstrating the feasibility of coating photocatalytic NPs with AMPs for selective boosting of antimicrobial effects: On coating TiO 2 NPs by positively charged LL-37 at pH 5.4-7.4,the peptide coating does not detrimentally interfere with ROS generation and displays good stability on UV exposure for 1-2 h.As a result of this, binding of peptide-coated TiO 2 is much higher than that of bare TiO 2 NPs, particularly at anionic bacteria-like membranes, and oxidative degradation of the latter on UV exposure strongly boosted by peptide coating, whereas corresponding effects on zwitterionic mammalian-like membranes were much smaller.Mirroring this, LL-37-coated TiO 2 NPs displayed boosted antimicrobial effects against both Gram-negative E. coli and Gram-positive S. aureus bacteria, whereas toxicity against human THP1 monocytes remained low.