Ink-Jet Printing-Assisted Modification on Polyethersulfone Membranes Using a UV-Reactive Antimicrobial Peptide for Fouling-Resistant Surfaces

Antimicrobial peptides (AMPs) are promising candidates for surface coatings to control biofilm growth on water treatment membranes because of their broad activity and the low tendency of bacteria to develop resistance to AMPs. However, general and convenient surface modification methods are limited, and a deeper understanding of the antimicrobial mechanism of action is needed for surface-attached AMPs. Here, we show a method for covalently attaching AMPs on porous ultrafiltration membranes using ink-jet printing and provide insight into the mode of action for the covalently tethered peptide RWRWRWA-(Bpa) (Bpa, 4-benzophenylalanine) against Pseudomonas aeruginosa. AMP-coated ultrafiltration membranes showed surface antibacterial activity and reduced biofilm growth. Fluorescence microscopy analysis revealed that the modified surfaces could cause cell membrane disruption, which was seen by live uptake of propidium iodide stain, and scanning electron microscopy images showed compromised cell membranes of attached bacteria. This study indicated that the mode of action of covalently tethered AMPs was similar to that of freely soluble AMPs. The deeper understanding of the mode of action of AMPs covalently attached to surfaces could lead to a more rational approach for designing surfaces with antibacterial activity.


INTRODUCTION
Membrane fouling is a major problem encountered in membrane filtration processes especially in applications such as wastewater treatment and desalination. Fouling in the pressure-driven membrane technology leads to increased economic and environmental costs because water permeation is reduced and chemicals are used to clean the surfaces. The use of chemicals not only reduces the membrane lifespan but also the chemicals used are ultimately released into the environment. Among various types of membrane fouling, biofouling (or microbial/biological fouling) is one of the major problems faced by the water treatment industries. Biofouling refers to any undesirable accumulation of living organisms on a surface causing a decrease in permeate production in a membrane filtration system. 1,2 Biofouling was termed the "Achilles heel" of membrane-based water technologies 3 because even if 99.9% of microorganisms are removed, the small number of remaining cells can multiply by colonizing on the surfaces and can grow by utilizing bioorganic substances in the feed water. Biofouling contributes to more than 45% of all membrane fouling, 4,5 and it is an inherent problem in reverse osmosis (RO), ultrafiltration (UF), and nanofiltration membrane filtration processes. An attractive strategy to delay biofilm formation is the fabrication of membrane surfaces with antibacterial properties, which might hinder the biofouling process. Inhibition of bacterial growth on the membrane surfaces might lead to effective separation processes and improved membrane performance. 6 UF membranes with antibacterial and antifouling properties might be advantageous because of their widespread use in food filtration processes and drinking water treatment.
UF membranes are widely used in industry and in a broad range of applications. More specifically, polyethersulfone (PES) membranes are extensively used in biomedical applications, water treatment, and other industrial fields. Various strategies have been explored to prevent or control biofilm growth on various surfaces, such as surface grafting of zwitterion or polymer brushes, antibiotic surface coatings, and the fabrication of new antifouling materials. 7−15 Antibiotic coatings such as silver are very effective at prevention of viable bacteria to proliferate on the surface; however, leaching of the silver ions into the solution leads to limited lifespan of the coating. A strategy that incorporates the antibacterial property into the material itself is the covalent attachment of antimicrobials such as antimicrobial peptides (AMPs) to the surface, which could mitigate the need to periodically re-apply the coating. 16 However, the retention of antimicrobial activity of immobilized AMPs is critical, and studies have showed that the antimicrobial activity of bound AMPs is generally found to be less compared to their soluble counterparts. 17,18 Thus, the fundamental understanding of antimicrobial activity of immobilized peptides is essential for developing efficient and long-lasting antimicrobial surfaces.
AMPs are peptides with broad spectrum activity that target microorganisms ranging from viruses to parasites. 19,20 Natural AMPs can be found in both eukaryotes (e.g., protozoan, fungi, plants, insects, and animals) and prokaryotes (e.g., bacteria). 21,22 In animals, AMPs are believed to be the first line of the innate immune defense 23−25 against fungi, viruses, and bacteria. For example, magainin is secreted from frog skin and thus is effective to prevent colonization of microorganisms on the skin's surface 26 and indicates that topical applications are suitable for such AMPs. Natural AMPs can be unsuitable for large-scale manufacturing because of the complexity or peptide length, which significantly increases the costs. Alternatively, short artificial AMPs have been proposed, and surprisingly short peptides have been shown to be very effective antimicrobial agents, for example, the RWRWRW sequence 27 and other designed lipopeptides that consist of even fewer amino acid residues but conjugated to hydrophobic compounds. 28 Despite the promising potential of these AMPs, there is limited information regarding surface immobilized AMPs and further investigation is needed.
Recently, the mode of action of the RWRWRW peptide and other related sequences was extensively studied and reported. 29,30 This peptide sequence was shown to target the lipid membrane, having no specific receptor−ligand interaction, but having the ability to interfere with multiple cellular processes. The cationic nature of the arginine residue and the interactions of the tryptophan residues with lipid membrane components can uniquely combine to associate with bacterial lipid membranes. 31 Similar artificial AMPs attached to surfaces might be valuable for suppression of bacterial growth because they would still be able to interact with bacterial membranes in multiple ways. 32 In order to investigate the antibacterial mode of action of immobilized AMPs, we attached the similar, yet augmented sequence RWRWRWA-Bpa [Bpa, 3-(4benzoylphenyl)alanine] on the PES membrane by a UV inkjet printing-assisted modification method. Benzophenone is a photoreactive compound that has been shown to react covalently with compounds at close proximity using ∼365 nm ultraviolet irradiation. 33,34 Incorporation of benzophenone in the peptide sequence and application on the surface with printing and direct UV irradiation can lead to a covalent attachment of the peptide onto surfaces. In our previous study, we used the same peptide sequence RWRWRWA-Bpa for modification of the surface of a RO membrane. 35 The peptide was concentrated on the surface by filtration after which UV irradiation covalently attached the peptide to the membrane. In the present study, the ink-jet printer was used to both apply the peptide on the surface and activate with UV. This new method increases the versatility of the coating process because porous surfaces such as UF membranes can now be modified in a process that combines the coating and UV activation in one step. Printing could also improve the precision of the process and reduce the consumption of chemicals and moreover opens new possibilities for modification such as patterned membrane surfaces. 36

RESULTS AND DISCUSSION
A new method for immobilization of AMPs on UF membranes was developed using an ink-jet printer with a UV light fixed to the print head, which applied the photoreactive peptide solution to the surface followed with immediate UV irradiation. In this way, multiple coatings can be printed in order to increase the peptide amount on the surface. To estimate the amount of peptide printed on the surfaces, we theoretically calculated and experimentally measured the volume of peptide solution printed on the membrane. Using the ink concentration of 3.33 mg/mL, the theoretical value of peptide printed on the membrane was 3.72 μg/cm 2 (37.2 mg/ m 2 ). This was compared to experimentally weighing the ink that was printed on a membrane, which gave a peptide amount of 20 ± 2 mg/m 2 . However, because such small volumes were printed, significant evaporation could have occurred in the time from printing to weighing the sample. Thus, the actual amount of printed peptide was probably between the measured and theoretical values.
2.1. Membrane Surface Characterization. UF membranes were printed using the UV-printing method (see Materials and Methods) with 0, 1, 2, or 4 layers of peptide ink and compared with membranes printed with ink only. Each sample was subjected to intense washing with sonication in order to ensure that only covalently bound RWRWRWA-(Bpa) remained on the membrane coupons. Fourier transform infrared spectroscopy (FTIR) characterization showed a new signal at 1660 cm −1 , which was probably due to the amide bonds of the attached peptide ( Figure 1a). 38 This signal was correlated with the amount of peptide printed, and as the number of printed layers on UF membranes increased, the signal also increased. This absorption signal was normalized to the PES membrane material using the ratio between the amide peak intensity and the membrane peak at 1487 cm −1 , 39 and this peak ratio corresponded to the peptide degree of grafting ( Figure 1b). Comparably, the peak ratios from the membranes printed with only the base ink were always less than the peptide printed membranes and did not show an increasing trend. This method is different than modification of UF membranes by photoassisted graft polymerization via direct UV excitation of the polymer, which employs acrylic monomers: 40 In the present method, the photoreactive amino acid 3-(4-benzoyl)phenylalanine 41 was incorporated into the peptide and is used to link the peptide covalently to the surface.
In order to verify that the peptide was indeed attached to the surface, visualization was done with the Ponceau S staining method, which can detect micrograms of peptide and shows a reddish-pink color when attached to the peptides. Ponceau S is a negatively charged dye and can bind to the positively charged amine and guanidine groups in the peptides. We observed that the areas of the membrane that underwent peptide modification were stained pink, whereas the unmodified membrane did not change color ( Figure 1c). Next, X-ray photoelectron spectroscopy (XPS) was performed, which also revealed the presence of the peptide. The atomic percentage of N 1s increased significantly from 1 to 5.6% after attachment of RWRWRWA-(Bpa) on the PES membrane support ( Figure  1d). Deconvolution of the N 1s peak at 400.5 eV showed specifically that the N−H bond from indole 42 and N−H from amide bonds were present and corresponded to the bound peptide sequence, which contained tryptophan (3 of 8 residues). Taken together, this evidence indicated that the UV ink-jet printing modification was successful.
The hydrophilicity of the membrane surface was moderately increased as more peptide was attached as seen from a decreasing water contact angle (CA) (Figure 1e). The CA decreased from 87°± 2.1 for the untreated membrane to 76°± 2.8 for the membrane printed with four layers of peptide. For the membrane that was printed only with the ink, a minor decrease was observed (83°± 0.7) and might have been due to the adsorption of the base ink components to the surface or slight degradation of the PES due to the UV treatment. 43 Although the difference between the peptide printed membranes was very minor, the differences between 1 and 2 prints or 1 and 4 prints were significant (P = 0.041 and 0.004, respectively). The peptide printing modification also affected the pure water permeability. After peptide modification, the membrane permeability decreased from 691 ± 76 LMH/bar for the control membrane to 360 ± 69 LMH/bar for the membrane printed with four layers of peptides and corresponded to a 48% flux reduction ( Figure 2). A possible reason for such a decrease could be the immobilization of peptide inside the small pores at the membrane surface, which could have reduced the average pore size in the barrier layer. Compared to an unmodified membrane, the energy requirement for filtration processes utilizing the modified membrane would be thus higher due to elevated pressures needed for the same water production. However, the resistance to membrane biofouling might lead to less frequent cleaning procedures and longer lasting membranes, which might mitigate the extra energy costs.
2.2. Bacterial Inhibition on the Surfaces. It has been previously demonstrated by us and by others that peptides can kill bacteria while attached to surfaces, 35,38,44,45 but less is known about the mode of action of tethered antimicrobial compounds. Thus, in order to confirm that these surfaces were antimicrobial, we performed a bacterial inhibition assay where the viability of a culture of Pseudomonas aeruginosa was measured after being contacted with the surface. A low-tomoderate killing effect was observed, and a maximum inhibition of 36% was seen on the peptide printed membrane (four layers of printed peptide). The inhibition decreased as peptide amount decreased (Figure 3a). The samples were also visualized using scanning electron microscopy (SEM) for a qualitative analysis of bacterial appearance (Figure 3b,c). In

ACS Omega
Article general, bacteria in contact with the UF membranes without AMPs showed smooth surfaces and uniform shape and appearance (Figure 3b). However, the appearance of bacteria after contact with the membranes containing tethered AMPs indicated stress, including structural damage (Figure 3c). Previous research showed that the peptide RWRWRW interacted with the cell membrane causing disruptions in many cell processes including, for example, delocalization of essential peripheral membrane proteins. 29 Our present observations indicate that the attachment of the similar peptide sequence RWRWRWA-(Bpa) on the surfaces of UF membranes can also cause disruption of the bacterial membrane and ultimately lead to an inhibitory effect, although disruptions in specific cellular processes were not studied.
Similarly, bacteria that were contacted with the membrane surfaces were visualized using fluorescence microscopy. SYTO 9 and propidium iodide were used to assess the viability of the cells. The dye SYTO 9 stains cells with intact cell membranes, whereas the bacteria with disrupted or damaged cell membranes are stained with propidium iodide. P. aeruginosa that was contacted to the membranes for 20 min was stained and transferred to a microscope slide and observed for 15 min (see Figure 4 and videos in the Supporting Information). Throughout the visualization period, the bacteria remained green and indicated that the bacteria were not adversely affected by contacting the unmodified membrane ( Figure  4a,b). In contrast, part of the bacteria population that were contacted with the peptide-coated membranes changed from green to red and indicated that the membranes became permeable to propidium iodide, which indicated membrane damage (Figure 4c,d). Bacteria samples that were treated with the free peptide in solution showed that the entire population of cells became permeabilized (Figure 4e,f). The similar action between the free peptide in solution and the tethered peptide on the surface indicated that the mode of action on the bacterial membrane is most likely the same. Because the peptide is covalently attached to the surface, the surface bound peptides are limited to interaction with the bacterial membrane and the relatively quick killing might indicate a detergent-like action.
2.3. Antibiofilm Activity. Antibacterial surfaces might lead to surfaces that delay biofilm growth because bacterial attachment is one of the first steps in biofilm development. Thus, we subjected the peptide-modified membrane surface to biofilm growth conditions in a flow cell experiment. The unmodified and modified surfaces were inoculated with P. aeruginosa and kept in a constant nutrient media flow for 96 hours to grow biofilm. Subsequently, confocal laser scanning microscope (CLSM) was used to quantify the biofilm. The peptide-modified membranes indeed showed a delayed biofilm growth, with a 66% reduction of biovolume and 32% reduction of average thickness after 96 h against P. aeruginosa when compared to the control-unmodified membranes ( Figure 5). In addition to the antimicrobial surface effect, the slightly increased hydrophilicity might have also contributed to delayed biofilm development.

CONCLUSIONS
A general membrane surface modification method was developed which utilized an inkjet printer fitted with a UV lamp, which allowed both photoreactive peptide deposition and immobilization on UF membranes. The modified surfaces showed antibacterial activity and delayed biofilm formation and appeared to maintain a membranolytic or detergent-like mode of action, which was limited to the bacterial membrane.

ACS Omega
Article This method could be applied for many purposes to a variety of polymer surfaces for surface functionalization with compounds that contain photoreactive linkers. To our knowledge, biofouling remains a significant challenge in membrane technologies and the concepts presented here to reduce biofilm growth and bacterial viability on surfaces can guide future research in developing alternative solutions to be used in reducing biofouling on UF membranes.

Membrane Modification with AMPs Using
Ink-Jet Printing. The Fujifilm Dimatix Material Printer (DMP-2800 series) was the ink-jet printer used and included a UV lightemitting diode (LED) (a wavelength of ∼365 nm, OmniCure LX500). The UV LED was purchased from Excelitas Canada Inc. and was fit onto the print head.
4.2.1. UV Printing Method. The UF membranes were first rinsed with 50% (vol) ethanol and then washed in a sonicator with DDW (10 min × 3 times). They were then stored in DDW at 4°C until used. Solutions of the benzophenyl-AMPs [RWRWRWA-(Bpa)] were made by dissolving 10 mg in 3 mL of C6A base ink. These solutions were passed through a 0.22 μm syringe filter to prevent any particles from clogging the cartridge or inner tubes or the ejecting nozzles, and added to the printer cartridge (10 pL drop volume). The printer cartridge volume was 3 mL. All printer nozzles (16) were open and printer settings were adjusted to obtain an even jetting velocity of 7−9 m/s. If multiple prints were performed, the time between prints was set to 5 min. The platen temperature was set to 50°C. Then, the UV lamp was attached to the print head and turned on, and the ink solution was printed using the setting of 30 μm drop spacing (DS) on a 40 mm × 40 mm square UF membrane. Afterward, the modified UF membranes were washed in DDW using sonication (3 × 10 min in an ice bath), dried, and were characterized and tested with FTIR and XPS, and the surface CA was measured. Average results including standard deviation are reported. The membrane samples were characterized by a VERTEX 70/ 80 spectrophotometer (Bruker Optiks GmbH, Ettlingen, Germany) with a MIRacle ATR attachment with a onereflection diamond-coated KRS-5 element. The IR spectrum was obtained with 40 scans at 4 cm −1 resolution, in a range of 400−4000 cm −1 using OPUS software (version 6.5) data management, averaging six replicates of random spots on the membrane surface. In particular, all of the modified membranes were completely dried at ∼25°C for 1 h before the analysis experiment.
4.3.1.1. X-ray Photoelectron Spectroscopy. X-ray photoelectron spectrometer ESCALAB 250 ultrahigh vacuum (1 × 10 −9 bar) installed with an Al Kα X-ray source (the beam size: 500 μm) and a monochromator was used for all of the membrane samples. The survey spectra were recorded with a pass energy (PE) of 150 eV, and high-energy resolution spectra were recorded with PE 20 eV. Also, utilizing nonlinear leastsquares curve fitting, different elements fixed the experimental data to the collected signals from C 1s, N 1s, and O 1s. To avoid charging effects, all spectra were corrected to the carbon 1s peak positioned at 285.0 eV. 4.3.1.2. Ponceau S Staining. Ponceau S staining solution consisted of 0.1% (w/v) Ponceau S in 5% (v/v) acetic acid. The modified and unmodified PES membranes were immersed in the staining solution (5 mL) and shaken on an orbital shaker for 5 min at room temperature. Next, each membrane was rinsed with distilled water (three times, 5 min each) to remove the excess dye, and the membranes were photographed.
4.3.1.3. Surface CA Measurement. The static CA was determined using the sessile drop method with deionized water by an OCA-20 CA analyzer (Data Physics Instruments, Filderstradt, Germany). Membranes were dried completely in the oven at 25°C for 1.5 h, and then a droplet (2 μL) was applied onto the surface. An image of the droplet was obtained

ACS Omega
Article immediately after it was dispensed on the membrane surface; then, using SCA-20 software (DataPhysics), the CA was calculated. Six replicate drops on a membrane sample were used to obtain the average for that sample, and three replicate sample membranes were measured. P-values were calculated using the t-test with 18 data points (6 measurements × 3 replicate membranes) for sample comparison.
4.4. Membrane Permeability. The water permeability of membranes was tested using a dead-end stirred cell (Amicon stirred cell50 mL, EMD Millipore Corporation, USA). Milli-Q water was added to the cell fitted with a membrane (42 mm diameter), and a pressure of 1.0 bar was applied. The permeate was collected and weighed every minute. An average of four measurements was used to calculate the membrane permeability (L p ) using where V (L) is the water volume collected, A (m 2 ) is the surface area of the tested membrane, t (h) is the elapsed time, and p (bar) is the applied pressure. 4.5. Contact Killing. A single P. aeruginosa colony from an LB agar plate was transferred to a 10 mL sterile LB broth and cultured overnight at 30°C. From this liquid culture, 100 μL was transferred to a 10 mL sterile LB broth and incubated for 9 h at 30°C on a shaker at 150 rpm. When the optical density at 600 nm was ∼1.0 (corresponding to ∼2.1 × 10 8 cfu/mL), the culture was serially diluted to 10 −4 of the original culture in sterile phosphate-buffered saline (PBS). The control membranes (unmodified PES membrane) and RWRWRWA-(Bpa)modified membranes were cut into square pieces (1 cm × 1 cm) and attached to a sterile glass slide using a double-sided tape and placed inside a Petri dish. Bacterial culture (50 μL) from the 10 −4 dilution (approximately 1000 cells) was placed on the surface of the membrane and evenly spread on the membrane by gently placing a sterile coverslip on top. The membranes were then incubated for 3 h at rt. After incubation, the modified and unmodified membranes and coverslips were washed with 500 μL of sterile PBS buffer.
4.5.1. Quantification Using Plate Count. Hundred microliters was spread on LB agar and incubated at 30°C overnight, and colony-forming units (cfu) were counted. 4.5.2. Visualization Using SEM. After the membrane surfaces were incubated with the bacteria culture, the coverslips were removed and 2.5% glutaraldehyde (100 μL) was added on the surface at 4°C and kept undisturbed overnight in a chemical fume hood at rt. The samples were washed twice with PBS, and 1% osmium tetraoxide (100 μL) was added on the surface and incubated for 1 h and then washed twice with PBS. Dehydration was carried out by immersion in a gradient of aqueous ethanol solutions (50, 70, 90, and 100%, v/v) for 15 min each and then tertiary butanol for 30 min. The dehydrated sample was air dried, coated with gold particles, and observed with SEM (Verios XHR 460L SEM).
4.6. Biofilm Assay. The biofilm growth assay 46 was performed as described previously. 35,47 Briefly, the AMPmodified UF membranes were cut into square pieces (1 cm × 1 cm) and attached to a cover slip (24 mm × 40 mm) by a double-sided tape. The samples were placed vertically in a flow cell, and utilizing a peristaltic pump, a culture of P. aeruginosa (50 mL, OD ≈ 0.1 at 600 nm at the exponential phase) was introduced. This was followed with sterile LB media (15 L) for 96 h at 0.44 cm/min cross-flow velocity. The membraneattached cover slips were then removed, and 100 μL of the staining solution [prepared by adding propidium iodide (1.5 μL, 20 mM) and SYTO 9 (1.5 μL, 3.34 mM) to an aqueous NaCl (0.997 mL, 0.1 M)] was added to cover the biofilm surface and stored protected from light for 20 min. Afterward, the membrane surfaces were gently washed (3×) using an aqueous solution of NaCl (0.25 mL, 0.1 M). Next, the sample membranes were observed under the CLSM (Zeiss LSM 510, META), with Zeiss dry objective plan-NeoFluar (20× magnification and numerical aperture of 0.5). The volume and average thickness of the biofilm were quantitatively analyzed by Matlab 2015b with proprietary algorithms, and the simulated biofilm images were generated from Imaris 3D imaging software (Bitplane, Zurich, Switzerland). 48 4.7. Fluorescence Microscopy. P. aeruginosa solution was prepared as described in the contact killing assay, and after being grown in LB to an optical density at 600 nm of ∼1.0, the bacteria were suspended in PBS for an OD 600 of 1.0. P. aeruginosa solution (100 μL) was placed on the surfaces of the modified membranes and the unmodified membrane. Also, to 100 μL of the P. aeruginosa solution, 10 μL of the peptide solution (1 mg/mL) was added in an Eppendorf tube and served as an internal positive control and all samples were incubated for 15−20 min at rt. After incubation, the membrane surfaces were washed with sterile saline water and to these solutions containing bacteria, 100 μL of staining solution containing propidium iodide and SYTO 9 was added (prepared as above in the biofilm assay) and immediately placed on a glass slide and covered with a clean cover slip for observation under a fluorescence microscope (Zeiss Axio Scope) using an excitation wavelength of 450−490 nm and a 520 nm cutoff emission filter. The samples were observed for 15 min. Images were obtained throughout the observation time for still images, which were also complied into a movie representing the entire observation period (see the Supporting Information) with AxioCam MRc digital cameras at magnifications of ×1000.