Antimicrobial Effect of Nitric Oxide Releasing Hydrogels on Staphylococcus Aureus Derived Proteases

The skin serves as a crucial barrier against environmental insults and invading pathogens. However, traumatic injury or skin disorder can compromise this barrier function, leasing to bacterial colonization and infection of the wound with microorganisms such as Staphylococcus aureus, which are normally present on healthy skin. The secretion of bacterial proteases such as V8 protease from S. aureus can disturb the equilibrium between extracellular matrix degradation and deposition during wound healing resulting in loss of barrier integrity. We report the feasibility of a nitric oxide (NO) releasing poly‐ε‐lysine (pεK) hydrogel to prevent loss of barrier function caused by V8 proteases. The fabrication and characterization of the pεK hydrogel and NO releasing properties in biologically relevant media are reported. The NO‐releasing pεK hydrogel have demonstrated bactericidal activity against a clinical isolate of S. aureus in complex physiological media and concurrently reduce the catalytic activity of secreted V8 protease. Moreover, pεK hydrogels are cytocompatible with keratinocytes and dermal fibroblasts. In contrast, Penicillin G loaded pεK hydrogels showed excellent antimicrobial efficacy but did not affect V8 catalytic activity. This demonstrates that NO‐releasing pεK hydrogels hold potential as an effective treatment for infected wounds reducing the microbial burden and inactivating bacterial secreted proteases.


Introduction
Healthy skin functions to protect the body form microbial contamination through a variety of mechanisms including providing a physical barrier. Skin disorders such as eczema, atopic dermatitis and seborrheic dermatitis can cause loss of skin the gram-negative bacterium, Pseudomonas aeruginosa and has demonstrated a strong correlation between infection severity and protease levels. [8] S. aureus is a highly pathogenic bacterial species and a prevalent nosocomial organism identified in dermal and soft tissue infections worldwide. [9] S. aureus strains secrete extracellular proteases, including V8 protease, aureolysin, staphopain A and staphopain B which can cause disruption of epithelial barriers, [10] impeding the healing rates of chronic wounds. [11] The production of these proteases can exacerbate S. aureus pathogenesis by disrupting the host immune response and can degrade the stability of the host tissue, thus increasing the severity of the infection. [12] Furthermore, the treatment of S. aureus infections is challenging as antibiotic resistant strains such as methicillin-resistant S. aureus (MRSA) are becoming more prevalent. [13] Consequently, there is a urgent need for developing novel antimicrobials with broad spectrum activity to combat antibiotic resistant S. aureus strains whilst mitigating the impact of bacterial proteases.
Nitric oxide (NO) is a potent antimicrobial agent and has a proven role in wound repair which makes it an excellent candidate for the treatment of wound infections. Much of the NO produced in skin occurs as a response to wounding and this takes place during the three stages of wound healing. [14] During normal healing, the production of NO has a very distinct temporal course with initially high concentrations which aid in inhibiting and clearing bacterial infection followed by lower levels allowing for the normal wound healing processes. [15] Consequently, a surplus of NO during wound healing, or overproduction at the wrong time point, may be just as detrimental as underproduction. NO's dual role in wound healing necessitates that an efficacious NO delivery platform be able to delicately balance NO's cytoregulatory and antimicrobial roles. [16] At sufficiently high concentrations (>1 µm), NO exhibits a bactericidal effect, whereas at lower concentrations NO act as a biodispersal agent, returning bacteria to a planktonic state. [17] Additionally, NO has been shown to interact with cysteine bacterial proteases which play a crucial role in replication and virulence of microorganisms responsible for disruption of immune function, making it of particular interest to healing infected wounds. [18] Various platforms including antimicrobial dressings, bandages and creams are utilized for treating infected wounds. Hydrogel-based dressings are a promising management strategy as their inherent hydrophilicity can provide moisture to the wound bed, facilitates nutrients and gases diffusion and have appropriate mechanical properties for handling. [19] Additionally, hydrogels can be loaded with active agents such as antimicrobials to combat infection and aid wound healing. [20] Peptide-based hydrogels, and in particular those fabricated with antimicrobial peptides offer an attractive proposition as they have an inherent antimicrobial component within the dressing material [21] Poly-ε-lysine (pεK) is such an antimicrobial peptide with demonstrable antimicrobial activity [22] and we have previously reported on the use of NO functionalized pεK peptide hydrogels as an effective vehicle for the controlled and sustained delivery of NO for the treatment of bacterial keratitis. [23] The aim of this study was to investigate the efficacy and feasibility of NO functionalized pεK hydrogels for the treatment of S. aureus infections in terms of lowering bacterial burden and maintaining skin barrier function by reducing bacterial protease activity. We report on the development of NO functionalized pεK hydrogel and the release in 2 biologically relevant media, broth and cell culture media. The release of NO under varying pHs [23] is also reported as the pH in wounds can differ based on infections and acute/chronic states. The mechanical properties of the hydrogels were investigated to ensure that the materials were within range for an ideal wound dressing. The impact of the pεK hydrogels on V8 protease catalytic activity and the biocompatibility of the hydrogels against keratinocytes and dermal fibroblasts were investigated. These hydrogel are compared to Penicillin loaded pεK hydrogels.

Diazeniumdiolate Functionalized Poly-ε-Lysine Hydrogels
The fabrication of a NO producing N-diazeniumdiolate functionalized pεK hydrogel involves a two-step process as previously reported [23,24] and is presented in Figure S1, Supporting Information. In the first step, pεK is crosslined with nonanedioic acid using N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide and N-hydroxysuccinimide chemistry. Previous studies have shown that a hydrogel with excellent physical and mechanical properties can be produced by crosslinking 0.1 g mL −1 pεK cross-linked with 60 mol% nonanedioic acid. [24,25] Alteration in polymer density, crosslinker length or degree of crosslinking. can be used to tune the water content, oxygen permeability, tensile and compressive modulus. [24,25] In the second-step amine/amide groups on the pεK backbone are functionalized by exposure to NO gas under high pressure and alkaline conditions to produce N-diazeniumdiolate functional groups (pεK/NO). These functional groups spontaneously decompose to release two moles of NO and regenerate the amine precursor with the rate of release dependant on the pKa of the amino precursor (i.e., the lower the pKa, the faster the release under acidic conditions). [23] Indeed our group, and others have demonstrated various N-diazeniumdiolates undergoing a burst release of NO at pH 4 and a longer sustained release profile at physiological or higher pHs. [26] Changes in pH occur during wound healing the pH shifting from weakly acidic (<pH 6) and physiological (≈pH 7.4) environments with differences also arising between acute and chronic wounds. [27] N-Diazeniumdiolate functionalization of the pεK gels has previously been verified using Fourier-transform infrared spectroscopy (FTIR) and measurement of NO release using a chemiluminescent NO analyzer. [23] In phosphate-buffered saline (PBS), at 22 °C and pH 7 the total NO release 100 µm. This consisted of a burst release of NO that peaked at a maximum of 4.54 ± 0.29 µm for 90 s followed by a steady release of 0.38 ± 0.04 µm for 2 h ( Figure 1A). Assessing NO release in complex media such as Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/ F12) at the same pH (7), has shown a dramatic reduction in the burst release of NO. This can be attributed to the presence of several metabolites such as riboflavin which can act as scavengers reducing the amount of freely available NO. [28] In complex media the release of NO showed a burst release of 0.21 ± 0.08 µm after 5 min, tapering toward 0.13 ± 0.04 µm after 40 min for a further 80 min. The influence of pH (4-8.5) over NO release over a short www.advmatinterfaces.de (45 min) period was also examined ( Figure 1B). The results here show that at a lower pH higher burst release of NO is observed and this decreases with increasing pH.
The compressive modulus, water content and oxygen permeability [29] of each hydrogel was determined ( Figure 1C-E). These experiments also included a control pεK hydrogel containing physisorbed penicillin G (pεK/PenG) with proven efficacy against methicillin susceptible S. aureus (ATCC 25 293). [23] Previous work has demonstrated penicillin G was attached to the pεk hydrogels via electrostatic interactions. High-performance liquid chromatography (HPLC) analysis was used to study the elution of PenG from pεk/PenG hydrogels over a 5 h time period. It was demonstrated that 95% of PenG released from pεk/PenG hydrogels within the first 30 min. The compressive modulus ( Figure 1C) of the hydrogels decreased from 2.25 ± 0.20 to 1.94 ± 0.28 to 1.79 ± 0.11 MPa for the pεK, pεK/PenG, and pεK/NO hydrogels, respectively. Although the functionalization of N-diazeniumdiolates significantly impacted the mechanical properties, the pεK/NO gels exhibited mechanical properties in the range acceptable for wound healing devices. The elastic modulus of human skin ranges between 4.5 and 8 kPa, [30] and human dermal fibroblasts proliferation has been observed to be directly proportional to the stiffness of the tested substrates (E = 0.5-120 MPa). [31] The water content decreased from 72.67 ± 1.10% to 70.24 ± 2.02% and to 64.19 ± 1.24% for pεK, pεK/PenG, and pεK/NO hydrogels, respectively. Finally, the oxygen permeability also decreased from 29.92 ± 1.31 to 27.22 ± 2.12 to 21.37 ± 1.04 Dk for the pεK, pεK/PenG, and pεK/NO hydrogels, respectively. The hydrogel with high water content and oxygen permeability were designed to hydrate wounds and used for wound healing applications. [32] Both pεk hydrogels had a high water content, where the commercial hydrogel typically range from 30% to 80% and can be classified as low (>40%), medium (40%-60%), and high (<60%) water content groups. [33] Water in the hydrogel can facilitate oxygen permeability and transport. [34] The range of oxygen permeability of commercial contact lenses hydrogel is between 6 and 150 dk/t.

Impact Against V8 Protease Catalytic Activity and Bactericidal Efficacy
Free pεK in solution has a previously reported antibacterial activity against S. aureus [35] due to its highly cationic nature as a result of an abundance of amine groups inciting lipid membrane destabilization by interacting with negatively charged phospholipid head groups. Therefore, to begin with we briefly confirmed the antimicrobial activity of our free pεK and pεK/NO in solution against planktonic S. aureus (ATCC BAA-977) in Luria-Bertani broth (LB broth) observing minimum inhibitory concentration (MIC) values of 180 µg mL −1 ( Figure S2, Supporting Information). N-diazeniumdiolate functionalization www.advmatinterfaces.de of free pεK in 0.5 m NaOH was verified by Ultraviolet-visible (UV-vis) spectroscopy with a characteristic shoulder at ≈252 nm [36] that upon the equimolar addition of HCl leads to rapid (<1 min) diazeniumdiolate decomposition. Simultaneously, the generation of 4 maxima between 340 and 390 nm is indicative of nitrous acid formation that occurs via the reaction of NO with water and oxygen, though these maxima subsequently begin to dissipate over a period of 6 h and are absent after 24 h [37] ( Figure S3, Supporting Information).
To investigate the bactericidal activity of our pεK hydrogels we investigated their activity against a wound clinically isolated strain of S. aureus (ATCC BAA-977) originally attained from a dermal wound with inducible clindamycin (lincosamide) resistance. Clindamycin is one amongst several antibiotic agents used to combat S. aureus infections in diabetic foot ulcers [38] with the detrimental emergence of numerous clindamycin resistant S. aureus isolates reported in various clinical environments. [39] We confirmed the constitutive secretion of V8 protease from S. aureus (ATCC BAA-977) in complex media ( Figure  S4, Supporting Information) as protease secretion is not ubiquitous and can vary significantly amongst strains. [40] The feasibility of the various gels to target V8 protease was evaluated against S. aureus (ATCC BAA-977) in complex media (LB broth and DMEM) at 4 and 8 h time points (Figure 2). Longer time points of analysis was not feasible as bacteria did not grow in cell culture reproducibly beyond this point Figure 2A demonstrates the effect of the various hydrogels against a commercially available synthetic source of V8 protease, Z-Leu-Leu-Glu-Amino-4-methylcoumarin (Z-LLE-AMC). [41] A 50% reduction in V8 protease (5 µg mL −1 ) catalytic activity was measured after a 16 h exposure to pεK/NO (39 mg mL −1 ) in complex medium compared to pεK and pεK/PenG where catalytic activity remained unaffected. This indicates that only NO has deleterious action against the V8 protease and that penicillin or control pεK gels do not affect this pathway.
The bactericidal activity of the pεK, pεK/PenG, and pεK/ NO hydrogels (45 mg) was evaluated relative to untreated controls (T 0 = 5 × 10 4 CFU mL −1 ) in LB broth and complex media (500 µL) after 4 ( Figure 2B-LB broth, 2C-DMEM/F12) and 8 h ( Figure 2D-LB broth, 2E-DMEM/F12) at 37 °C/180 rpm. In LB broth after 4 h ( Figure 2B), the pεK/NO displayed a 3 log reduction in the bacterial cell counts in comparison to the bacterial control. The pεK/PenG gels displayed complete kill of S. aureus at this same time point. After 8 h, in LB broth ( Figure 2D), regrowth of the bacteria is observed, with pεK/NO gels demonstrating a 1.5 log reduction and pεK/PenG gels a 3 log reduction from the bacterial control. In addition, no significant difference has been found in the statistical analysis of factor of time (4 and www.advmatinterfaces.de 8 h) on material groups of untreated, pεK/PenG and pεK/NO, where pεK gels presented significances. In DMEM/F12 at 4 h ( Figure 2C), the pεK/NO gels and the pεK/PenG gels displayed a 0.5 log reduction and complete kill, respectively, compared with the bacterial control. After 8 h in DMEM/F12 ( Figure 2E), the pεK/NO and the pεK/PenG gels displayed a no statistical difference and 3 log reduction, respectively, in comparison with the bacterial controls. Again, no significant difference has been found in the statistical analysis of factor of time (4 and 8 h) on material groups of pεK/PenG and pεK/NO, where untreated and pεK gels displayed significant differences. The diminished antimicrobial activity of the Pεk/NO in DMEM is corroborated by the chemiluminescence results presented in Figure 1A. NO is quenched by scavengers such as riboflavin in DMEM/ F12 media thereby reducing the overall concentration that is available for antimicrobial function. These data indicate that although pεK/PenG has better efficacy against S. aureus overall, the mechanism of NO action is partially effective via inactivation of the V8 protease.
S. aureus secretes the bacterial protease V8 that is responsible for increasing the virulence of the bacteria. [42] Specifically, the V8 protease hinders the immune response through the degradation of antimicrobial peptides such as LL-37 [43] and has an adverse effect on wound healing by perturbing the equilibrium between extracellular matrix deposition and degradation, contributing to the formation of chronic wounds. [44] Therefore, using an antimicrobial that targets the V8 bacterial protease could be beneficial and offers the possibility of lowering the virulence of the infection.

Biocompatibility Against Human Dermal Keratinocytes and Human Dermal Fibroblasts
The effect of our pεK, pεK/PenG, and pεK/NO hydrogels on keratinocyte (HaCaT) and dermal fibroblast (WS1) human cell lines (Figure 3A,B) was measured via extract and indirect testing as per ISO10993-5. The hydrogels were incubated in cell culture media (DMEM/F12) for 24 h at 37 °C/5% CO 2 to produce a conditioned complex media. Each cell type was then challenged with conditioned complex media for 24 h at 37 °C/5% CO 2 prior to analysis of cell metabolic activity via an alamar blue [45] assay and cell metabolic activity with a lactate dehydrogenase (LDH) [46] assays. Minor reductions were only observed in cell metabolic activity for pεK/NO against HaCaT cells and for pεK, pεK/PenG, and pεK/NO against WS1 cells ( Figure 3C) with each yielding a slight but statistically significant decrease, although these values were still ≈90% or higher of untreated controls, and thus the gels are deemed non-cytotoxic according to ISO10993-5. Cell membrane integrity assays  Figure 3D) demonstrated no statistical differences and therefore cytotoxicity in the cells exposed to pεK, pεK/PenG, or pεK/ NO compared to untreated cells and therefore, as a result of all these tests each hydrogel was deemed acceptable for further investigation.
A more stringent indirect test of our pεK, pεK/PenG, and pεK/NO hydrogels was also performed. In brief, this involves a monolayer of cells separated (≈1 mm) from pεK, pεK/PenG, or pεK/NO via a cell culture insert with a polycarbonate membrane (8 µm pore size), with the relevant hydrogel placed on the apical membrane surface. This allows the rapid diffusion, interaction and assessment of any released molecules (including NO) without physical aberration of the intact monolayer upon application and removal of the hydrogel. Under these conditions no reductions in cell metabolic activity ( Figure 3E) and cell membrane integrity ( Figure 3F) compared to untreated controls was detected with either cell type. Each hydrogel therefore exhibits suitable cytocompatibility with each cell type under all the conditions examined.

Conclusions
The primary role of healthy skin is to provide an effective barrier against a wide range of invading microorganisms. However, injury to the skin can create a wound site that is susceptible to infection, which can impede the wound healing process. Therefore, the treatment of wounds should consider a two-pronged approach that involves reducing the microbial burden and facilitating wound repair. In this paper, we have reported on the development of a NO-releasing pεK hydrogel that is effective in reducing the microbial load and maintaining skin barrier function. We have investigated the release of NO in various biological solutions to ensure the robustness of our methodology, considering that bacteria thrive in broth while cells grow better in cell culture media. Our findings indicate that the NO-releasing pεK hydrogel has significant antimicrobial efficacy against a clinical strain of S. aureus. Furthermore, these hydrogels have the capacity to reduce the activity of bacterial-secreted V8 protease, which can negatively impact skin barrier function. In contrast, Penicillin G loaded pεK hydrogels showed excellent antimicrobial efficacy but did not affect V8 catalytic activity which can deteriorate skin barrier function. These results demonstrate that the NO-releasing pεK hydrogel is a promising dual-functional antimicrobial and wound repair treatment platform for infected wounds. Our findings suggest that this novel approach could provide an effective strategy for the treatment of infected wounds and ultimately improve patient outcomes.

Experimental Section
Materials: All materials unless otherwise specified were purchased from SigmaAldrich Company Ltd (Poole, UK) or ThermoFisher Scientific Ltd (Loughborough, UK) and were of the highest available purity.
Bacterial Propagation and Cell Culture: S. aureus (ATCC BAA-977) was purchased from the national collection of type cultures (strain designation/ NCTC 13 811), Public Health England (Porton Down, UK). HaCaT cells were kindly donated by Hamill lab (University of Liverpool, UK), WS1 (ATCC CRL-1502) cells were purchased from the American Type Culture Collection (Manassas, VA, USA). S. aureus was propagated where described in Luria-Bertani (LB) broth or LB-agar (lennox modification). HaCaT and WS1 cells were cultured in complex media, defined as DMEM/F12 containing phenol red supplemented with 10% (v/v) FBS and 100 U mL −1 Penicillin and 10 µg mL −1 Streptomycin. Complex media without FBS was used in chemiluminescence experiments due to the surfactant action of serum proteins and complex media without 100 U mL −1 Penicillin and 10 µg mL −1 Streptomycin was used in experiments with S. aureus. pεK, pεK/NO, and pεK/PenG Hydrogel Fabrication: To yield a 0.1 g cm −3 60 mol% crosslinked gel, pεK (Zhengzhou Bainafo Bioengineering Ltd, China) monomers with a molecular weight (MW) of 3.2-4.5 kDa (degree of polymerization [25][26][27][28][29][30][31][32][33][34][35] were dissolved to a concentration of 0.2 g mL −1 in an aqueous solution comprising 3.3% (v/v) N-methylmorpholine (NMM)), 1% (v/v) Tween 20 and 0.033 g mL −1 nonanedioic acid. To initiate gelation an equal volume of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDCI) and N-hydroxysuccinimide (NHS) at 5 and 1.5 molar equivalents in ddH 2 O was added at room temperature and the mixture left to polymerized overnight. Thereafter gels were washed fivefold in ddH 2 O for 5 min per wash followed by a fivefold wash sequence in 10% (v/v) NMM for 5 min per wash prior to a final fivefold wash sequence in ddH 2 O for 5 min per wash. Gels were then UV-irradiated and stored at 4 °C in ddH 2 O until required. For diazeniumdiolate functionalization pεK hydrogels disks (6 mm dia.) were placed in a 0.5 m NaOH solution within a stainless-steel reactor pressurized to 5 bars with NO for a period of 72 h to yield diazeniumdiolate functionalized primary and secondary amine groups. Upon removal gels were then stored at 4 °C in 0.5 m NaOH until required. For Penicillin G functionalization pεK hydrogels were immersed in 12.5 mm Penicillin G in ddH 2 O for 3 h. Excess Penicillin G was then removed with a threefold wash sequence in ddH 2 O for 5 min per wash. Gels were then stored at 4 °C in ddH 2 O until required.
Nitric Oxide Detection Via Chemiluminence: NO release from pεK/ NO hydrogels utilized a Sievers 280i Chemiluminescence NO analyzer (NOA/ Boulder, CO). Prior to analysis, the NOA was calibrated according to the manufacturer's protocol using NO standard gas (94.5 ppm) and air passed through a NO zero filter (0 ppm of NO). 2 discs (6 mm dia./ 65 mg/disc) were immersed in 5 mL of solution at 22 °C and purged with nitrogen gas at a flow rate of 70 mL min −1 to carry released NO from the solution to the analyzer. Additional nitrogen flow was supplied to the flask to match the collection rate of the instrument with 3 independent repeats reported.
Mechanical and Physical Testing of pεK, pεK/NO, and pεK/PenG Hydrogels: Unconfined compression testing utilized a Univert mechanical tester (CellScale Inc., Waterloo, Canada) equipped with a 50 N load cell. Gels of diameter 11.5 mm and thickness 2 mm were subjected to unconfined compression with 75% displacement over 60 s (or maximum force (50 N)) at room temperature. The resulting stress-strain curves were plotted with the young's modulus (Equation (1) For water content (by mass) analysis, pεK hydrogels (Dia. ≈8 mm/ Thickness ≈500 µm) were blotted to remove any excess fluid and weighed to establish their fully hydrated (wet) weight. Samples were then lyophilized and reweighed at regular intervals until no further loss in mass was detected. % water content (by mass) of 5 independent measurements was then calculated as shown in Equation (2).

% Water content by mass 100%
Wet mass mg Dry mass mg Wet mass mg Oxygen permeability (Dk) was calculated from the % Water content (by mass) values of 5 independent measurements attained using Equation (3). [29] Oxygen Permeability (Dk) 1.67e 0.397 % Water content (by mass) = ( ) * www.advancedsciencenews.com www.advmatinterfaces.de V8 Protease Catalytic Activity: The catalytic activity of V8 protease obtained commercially (New England Biolabs (UK) Ltd, Herts, UK) or in complex media conditioned with S. aureus (ATCC BAA-977) was assayed against an endoproteinase-GluC specific synthetic tripeptide Z-Leu-Leu-Glu-Amino-4-methylcoumarin (Z-LLE-AMC) (Enzo Life Sciences (UK) Ltd, Berkshire, UK) that upon proteolytic cleavage yields the fluorophore, Amino-4-methylcoumarin. In brief, 160 µL of sample and 20 µL of 10 000 U mL −1 Penicillin and 10 mg mL −1 Streptomycin was aliquoted in triplicate prior to the addition of 20 µL of 600 µm Z-LLE-AMC with V8 protease free, sample free and Z-LLE-AMC in ddH 2 O samples included as negative controls, within a flat-bottom black 96-well plate. Samples were then incubated at 37 °C for a period of 24 h prior to fluorescence measurements at excitation and emission wavelengths of 360/40 and 460/40 nm, respectively using a Synergy HT microplate reader (BioTek Inc., VT, USA.) with three independent repeats reported each containing multiple biological and technical replicates.
Bactericidal Activity of pεK, pεK/NO, and pεK/PenG Hydrogels Against S. aureus (ATCC BAA-977): The antimicrobial efficacy of pεK, pεK/NO, and pεK/PenG hydrogels was observed as the number of colony forming units (CFU) of S. aureus (ATCC BAA-977). In brief, an overnight culture in LB broth was diluted as required to 5 × 10 4 CFU mL −1 into pre-warmed medium. Thereafter samples were supplemented with 45 mg of the relevant pεK hydrogel and incubated at 37 °C/180 rpm. After 4 or 8 h a serial dilution in LB broth or complex medium (without 100 U mL −1 Penicillin and 10 µg mL −1 Streptomycin) as required was performed on LB agar using the Miles and Misra method [47] with three independent repeats reported each containing three biological replicates that each contained five technical replicates.
Cytocompatibility Testing: HaCaT and WS1 cells were seeded overnight at 3.33 × 10 4 cells cm −2 in either 96-well (extract) or 12-well (indirect) plates. For extract testing a pεK hydrogels (0.45 mm dia.) were incubated in triplicate in 1 mL of complex medium for a period of 24 h at 37 °C / 5% CO 2 to produce (extract) conditioned complex medium. Cells were then challenged for a period of 24 h prior to testing. For indirect testing 1.13 cm 2 polycarbonate membranes (pore size 8 µm/ 1 × 10 8 pores cm −2 , Merck Life Science, Gillingham, Dorest, UK) were situated 1 mm above the cell surface with pεK hydrogels (0.45 mm dia.) placed on the apical membrane surface with 900 and 100 µL of complex medium applied to the lower and upper compartments and samples incubated (37 °C/ 5% CO 2 ) for a period of 24 h prior to testing. Cytotoxicity testing utilized cell metabolic activity or cell membrane integrity assays. Cell metabolic activity assays were based on a modified alamar blue assay. [45] A 100 µL (extract) or 500 µL (indirect) aliquot of pre-warmed resazurin solution in complex medium was incubated with cells at 37 °C / 5% CO 2 for 4 h. Thereafter an 80 µL aliquot was placed in a black 96-well fluorescence plate and fluorescence measured using a Synergy HT microplate reader (BioTek Inc., VT, USA.) at excitation and emission wavelengths of 530/25 and 590/20 nm, respectively with values normalized between cell-free (−ve) and untreated (+ve) controls. Cell membrane integrity measurements were quantified by measuring LDH content [46] from intact cells and membrane compromised cells via a Pierce LDH Cytotoxicity Assay (Fisher Scientific UK Ltd, Loughborough, UK) and following the as supplied instructions.
Statistical Testing and Data Handling: All data handling and statistical analysis was performed using OriginPro 2015 (64-bit) sr2 (OriginLab Corp., MA, USA). Where appropriate one-way ANOVA tests (p ≤ 0.05) were used with specific post-hoc statistical tests and sample sizes stated in the corresponding figure legends.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.