Hyperbranched poly(NIPAM) polymers modified with antibiotics for the reduction of bacterial burden in infected human tissue engineered skin

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/47382487 Hyperbranched poly(NIPAM) polymers modified with antibiotics for the reduction of bacterial burden in... Article in Biomaterials · October 2010 DOI: 10.1016/j.biomaterials.2010.08.084 · Source: PubMed CITATIONS READS 29 160 6 authors, including: Joanna Shepherd Prodip Sarker 17 PUBLICATIONS 543 CITATIONS 14 PUBLICATIONS 143 CITATIONS The University of Sheffield SEE PROFILE Dr. Reddy's Laboratories SEE PROFILE Stephen Rimmer Sheila Macneil 160 PUBLICATIONS 2,325 CITATIONS 456 PUBLICATIONS 7,904 CITATIONS The University of Bradford SEE PROFILE The University of Sheffield SEE PROFILE All content following this page was uploaded by Stephen Rimmer on 12 December 2016. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document and are linked to publications on ResearchGate, letting you access and read them immediately. This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Biomaterials 32 (2011) 258e267 Contents lists available at ScienceDirect Biomaterials journal homepage: www.elsevier.com/locate/biomaterials Hyperbranched poly(NIPAM) polymers modified with antibiotics for the reduction of bacterial burden in infected human tissue engineered skin Joanna Shepherd a, b, Prodip Sarker c, Stephen Rimmer c, Linda Swanson c, Sheila MacNeil b, Ian Douglas a, * a Unit of Oral & Maxillofacial Pathology, School of Clinical Dentistry, University of Sheffield, Claremont Crescent, Sheffield S10 2TA, United Kingdom Department of Engineering Materials, Kroto Research Institute, North Campus, University of Sheffield, Broad Lane, Sheffield S3 7HQ, United Kingdom c Department of Chemistry, University of Sheffield, Sheffield S3 7HF, United Kingdom b a r t i c l e i n f o a b s t r a c t Article history: Received 29 July 2010 Accepted 27 August 2010 Available online 8 October 2010 The escalating global incidence of bacterial infection, particularly in chronic wounds, is a problem that requires significant improvements to existing therapies. We have developed hyperbranched poly(NIPAM) polymers functionalized with the antibiotics Vancomycin and Polymyxin-B that are sensitive to the presence of bacteria in solution. Binding of bacteria to the polymers causes a conformational change, resulting in collapse of the polymers and the formation of insoluble polymer/bacteria complexes. We have applied these novel polymers to our tissue engineered human skin model of a burn wound infected with Pseudomonas aeruginosa and Staphylococcus aureus. When the polymers were removed from the infected skin, either in a polymer gel solution or in the form of hydrogel membranes, they removed bound bacteria, thus reducing the bacterial load in the infected skin model. These bacteria-binding polymers have many potential uses, including coatings for wound dressings. Ó 2010 Elsevier Ltd. All rights reserved. Keywords: Bacteria Hydrogel Infection Microbiology Polymerization Wound dressing 1. Introduction Bacterial infection of burns, wounds and ulcers of the skin is a worldwide challenge, which is increasing with the rise of diabetes and advancing age of the population. Furthermore, the rise in bacterial antibiotic resistance is compounding difficulties of wound management [1]. There is a clear need, therefore, for improving management of both chronic bacterial skin infections and those which may occur in areas or situations remote from specialised clinical care, such as on the battlefield. Chronic infections of ulcers and pressure sores, which may be compounded by poor circulation and malnutrition, can delay wound healing for several years. Traditionally, topical antibiotics have been applied to infected wounds and the wounds dressed and re-dressed until the infection has diminished. This approach avoids systemic administration of antibiotics and the associated side-effects. However it can contribute to antibiotic resistance particularly when used in the treatment of long-term recalcitrant infected wounds. Alternatives to antibiotics have become more widely used because of this problem. For example, various silver compounds have been used, which although effective as antibacterial agents, can be toxic to the host tissue at high local * Corresponding author. Tel.: þ44 114 2717957; fax: þ44 114 2717894. E-mail address: i.douglas@sheffield.ac.uk (I. Douglas). 0142-9612/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2010.08.084 concentrations and treatment can have the effect of delaying wound healing [2]. In an effort to overcome some of these problems we have developed stimulus responsive hyperbranched poly (N-isopropyl acrylamide) (HB-PNIPAM) polymers conjugated with the antibiotics Polymyxin-B (HB-PNIPAM-pmx) [unpublished data] and Vancomycin (HB-PNIPAM-van) [3], that are able to selectively bind Gram-negative and Gram-positive bacteria respectively. In both cases the conjugated antibiotics do not have bactericidal activity but they bind bacteria via the latter’s target surface groups. PNIPAM has garnered increasing interest for use in biological systems, since as well as being non-toxic to host tissue, it has thermosensitive properties. At temperatures below 32  C, the polymer is hydrophilic and remains in a coiled form in solution. However, on raising the temperature past 32  C, which is its lower critical solution temperature (LCST), the polymer desolvates and switches to a globular formation. Since the first described interaction of biomolecules with PNIPAM in 1992 [4], this property has been exploited in the design of several biological agents [5e7]. Functionalization of the NIPAM monomer with other molecules can further modify the sensitivity of PNIPAM to the environment. For example, addition of acrylic acid or methacrylic acid results in microgels that are pH responsive [8], which have several potential uses including drug delivery. However, prior to our work, there have been no studies describing the transitions from coil-globule on binding to a cellular target. With HB-PNIPAM this is possible because the location of the ligands at chain ends ensures that they Author's personal copy J. Shepherd et al. / Biomaterials 32 (2011) 258e267 are available for binding both above and below the LCST: chain end groups tend not to penetrate the coil of the highly branched polymer [9]. Since the binding interactions involve a large perturbation of the overall solvation of the polymer, the LCST is reduced. The reduction of the LCST is observed as a coil-to-globule transition if the magnitude of the change is sufficient to reduce the LCST to below ambient temperature. The adhesion of bacteria to PNIPAM above the LCST and release below the LCST is well-known [10] and in our previous work [3] we showed that binding followed by the coil-to-globule transition is accompanied by aggregation of target bacteria. Linkage of antimicrobial peptides and antibiotics to other materials, such as titanium surfaces for surgical implants [11], sepharose beads for LPS decontamination [12], cloths/fibres [13,14] and membranes [15] have been investigated, but none of these materials possess the inherently useful responsive properties of PNIPAM. We have recently shown that at physiologically relevant temperatures, HB-PNIPAM-van in solution specifically binds Gram-positive bacteria, an event which induces a coil-toglobule transition resulting in the formation of an insoluble complex of polymer-bound bacteria [3]. We reasoned therefore, that our polymers may be able to act as a dressing to actively remove bacteria from infected tissue. To evaluate the usefulness of these polymers in a simulated in vivo environment, we recently developed a three dimensional tissue engineered model of bacterial infected burnt human skin. This was infected with either the Gram-positive bacterium Staphylococcus aureus or the Gram-negative bacterium Pseudomonas aeruginosa [16]. In the current study, we used this human infected skin model to examine the ability of both the free polymers and polymers bound to hydrogel membranes to selectively bind and remove P.aeruginosa or S.aureus from these infected wound beds. 2. Materials and methods 2.1. Materials HB-PNIPAM-COOH was prepared as previously described [17]. N,N-dimethylformamide (DMF) (Aldrich, sure-seal) and dioxane (Aldrich, sure-seal) were used as received. N,N-dicylohexylcarbodimide (Aldrich, 99%) and N-hydroxy Succinimide (Aldrich, 98%), 1-Amino anthracene (Aldrich) were used as received. IPA (isopropyl alcohol) and Ethanol (Fisher) HPLC grade were used. TriseHCl was used as purchased. Polymyxin-B Sulphate (Fluka), Vancomycin hydrochloride (SigmaeAldrich), polymyxin acylase as enzyme (Sigma), Fmoc-Chloride (Fluka, >98%), piperidine (Aldrich, 99%) were used as supplied. Ethylene glycol dimethacrylate (EGDMA) was passed through a column of basic alumina to remove the inhibitor. Glycidyl methacrylate (GME) was distilled under reduced pressure prior to use. Ethylene diamine was obtained from Fluka (UK) was used without further purification. Photo initiator, 2-hydroxy-2-methylpropiophenone (Aldrich) was used as received. 2.2. Modification of polymyxin-B sulphate FMoC-pmx was prepared in aqueous medium. A mixture of fluorenylmethyloxycarbonyl chloride (200 mg) and 1M Na2CO3 aqueous (5 cm3) was added to a solution of polymyxin-B sulfate (500 mg in 10 cm3 water) and this was stirred at room temperature over-night. FMoC-pmx was insoluble in water and it was filtered off and washed with water several times to remove unreacted reagents. In FTIR spectra, new bands due to FMoC (9-Fluorenylmethoxycarbonyl) group appeared at 1450 cm 1 and from 1300 cm 1e1200 cm 1. FMoC-pmx had a melting point of 236  C and the structure was confirmed by mass spectrometry (observed m/z 2314 Da (FAB-MS), theoretical exact mass 2313.090). To remove the terminal alkyl chain, a mixture of FMoC-pmx (50 mg in 16 cm3 50 mM TriseHCl, 0.3945 g TriseHCl in 50 cm3 water) and polymyxin acylase (2 cm3, 16 mg polymyxin acylase in 16 cm3 phosphate buffer pH 8.0) was stirred at room temperature over-night. Since the deacylated product (FMoC-pmx-deacylate) was insoluble in water and common organic solvents, it was filtered and washed several times to remove the unreacted reagents. In FTIR, a band at 1100 cm 1 due to the alkyl chain present in the FMoC-pmx was absent after enzymatic cleavage. The melting point was 277  C and the structure was confirmed by mass spectrometry (observed m/z 2174 Da (FAB-MS), theoretical exact mass 2172.97). 259 2.2.1. HB-PNIPAM-COOH with 1-amino anthracene at a fraction of the chain ends (HB-PNIPAM-A-suc) HB-PNIPAM-COOH (70 mg) was dissolved in DMF (1.5 cm3) and N,N-dicylohexylcarbodimide (DCC) (17 mg) and N-hydroxy Succinimide (9.5 mg) in DMF (1 cm3) were added to it. The solution was stirred under N2 over-night and solvent was removed by rotary-evaporator. The solid was dissolved in acetone/ethanol mixture (9:1) and concentrated by ultra-filtration over 1 h. The ultra-filtration was repeated three times; solvent was removed by rotary-evaporation and dried under vacuum at room temperature. The solid was dissolved in DMF and precipitated in diethyl ether, polymer filtered off and dried under vacuum at room temperature. HB-PNIPAM-succinimide, 300 mg was dissolved in DMF (10 cm3) and a solution of disopropylamine (3 mg) and 1-aminofluorescene (3 mg) in DMF (6 cm3) were added to it. The reaction mixture was stirred at room temperature over-night. The solvent was removed by rotary-evaporation and then the solid was dissolved in acetone/ethanol mixture (9:1) and concentrated by ultra-filtration over 1 h. The ultra-filtration was repeated three times; solvent was removed by rotary-evaporation and dried under vacuum at room temperature. The solid was dissolved in DMF again and precipitated in diethyl ether, filtered off and dried under vacuum at room temperature. 2.2.2. Synthesis of HB-PNIPAM-van and HB-PNIPAM-pmx HB-PNIPAM-A-suc (50 mg) was dissolved in water (2.5 cm3) over ice. Then the solution transferred to a small reaction tube. A solution of Vancomycin (15 mg) in water (1 ml) and 0.1 M sodium phosphate buffer (pH 8.5) (1 cm3) were added to the polymer solution. pH of the total mixture final mixture was maintained at 9.5 and stirred over-night on ice. The solution was ultra-filtered three times, freeze-dried and stored at 20  C. For coupling between activated HB-PNIPAM-A-suc and FMoC-pmx-deacylate-B DMF was used as a solvent and the reaction mixture stirred for two days at RT. 20% piperidine aqueous solution was added during ultra-filtration to remove the FMoC groups. The solution was ultra-filtered three times using de-ionized water and then freeze-dried and stored at 20  C. 2.2.3. Photopolymerisation Crosslinked hydrogel membranes were prepared as follows. The monomers, NIPAM (6.75 g), glycidyl methacrylate (GME) (0.45 g) and ethandiol dimethacrylate (EDMA) (1.8) were dissolved in Dioxane (5 ml) along with 2-hydroxy-2-methylpropiophenone (90 mg, 1 wt.% of total monomers). The mixture was added to a polymerization mould and irradiated with a 200 arc 400 W mercury discharge lamp at a distance of 10 cm in a Dimax model Bondbox on a rotating table for 2 min on each side. The polymerization mould consisted of two 4 mm thick quartz glass sheets covered with 100 mm PET film attached by the minimum amount of spray-mount adhesive (3 M). The plates were separated with a rectangular 500 mm PTFE spacer and the sheets were 125 mm  70 mm  5 mm when removed from the mould. 2.2.4. Modification of the membrane Amine-functional membranes (10 g) were produced by soaking freshly prepared membrane in ethanolic solution containing the 1,2-ethandiamine (250 ml, 5 vol.%) in large excess for 24 h at room temperature. The operation was conducted in screwtop polypropylene bottles and membranes were washed several times with ethanol and then IPA to remove excess diamine. HB-PNIPAM-van and HB-PNIPAM-pmx with residual succinimide end groups were prepared as described below and the residual succinimide groups were then reacted with the amine-functional membranes. Functionalized HB-PNIPAM-van (100 mg) was dissolved in 20 cm3 water (this was carried out over ice for 30e40 min), then the solution was transferred to a screw-top polypropylene bottle. Pieces of amine-functional membranes (5 g) were added to it, and then 5 ml of 0.1 M sodium phosphate buffer (pH 8.5) was added to the solution. The pH of the final mixture was maintained at 9.5 and the mixture was stirred overnight over ice. The modified membranes were washed with water and IPA respectively several times and finally stored in IPA. For coupling between functionalized HB-PNIPAM-pmx and amino functionalized membranes, IPA was used as a solvent and the reaction mixture was stirred for two days at RT. 2.2.5. Synthesis of HB-PNIPAM-van or -pmx containing residual succinimide end groups Succinimide activated polymer (HB-PNIPAM-suc) (500 mg) was dissolved in water (10 cm3) over ice and then Vancomycin (24 mg) in water (2 cm3) and sodium phosphate buffer (0.1 M, pH 8.5, 5 cm3) were added. The pH of the mixture was maintained at 9.5 and the mixture was stirred at 2e0  C over-night. The solution was ultra-filtered three times and then freeze-dried and stored in a freezer. For coupling between activated HB-PNIPAM and FMoC-pmx-deacylate, DMF was used as a solvent and the reaction mixture was stirred for two days at RT. 20% piperidine aqueous solution was added during ultra-filtration to de-protect the amine groups. The solution was ultra-filtered three times using de-ionized water to remove DMF and unreacted reagents and then freeze-dried and stored at 20  C. 2.2.6. Swelling of the membranes The equilibrium water contents (EWC ¼ wet weight-dry weight/wet weight  100) were determined gravimetrically in de-ionized water on cylindrical Author's personal copy 260 J. Shepherd et al. / Biomaterials 32 (2011) 258e267 samples cut using a cork borer. 6 sample disks per material were cut with a number 2 cork borer (6 mm diameter) and were hydrated over 12 h in ultrapure water. The equilibrium water contents of the membranes at various stages were: EWC (wt%) As polymerized membrane Amine-modified VAN-functional PMX-functional 26 27 33 34 2.2.7. Bacteria Clinical isolates of S.aureus (S-235) and P.aeruginosa (SOM-1) were used as representative Gram-positive and Gram-negative strains respectively. Bacteria were cultured under standard conditions on brain heart infusion (BHI) agar (Oxoid), and in BHI broth for experimental use. Following over-night incubation in broth, bacteria were washed and resuspended in PBS to a concentration of 1  1010 cfu/ml. 2.2.8. Tissue engineered skin constructs Constructs were produced as previously described [16]. Briefly, human dermal fibroblasts and keratinocytes were isolated from split-thickness skin biopsies received from abdominoplasties and breast reductions performed at the Northern General Hospital, Sheffield. Research Ethics approval was obtained from the Sheffield Research Ethics Committee. Decellularised dermis (DED), (produced by removing cells using 1 M NaCl as detailed in [16]) also sourced from human skin biopsies, was used as a base scaffold. Rings of DED 15 mm in diameter were cut and placed within 12 mm tissue culture inserts with 4 mm pores in the base (Greiner). Inserts were suspended from the edges of 12 well plates into the wells. Greens’ medium containing 10% foetal calf serum was added to the bottom of the wells so that it lapped the under surface of the DED. The DED was then seeded with 1  105 fibroblasts and 5  105 keratinocytes, each in 250 ml of 10% Greens medium. Again, after 24 h incubation at 37  C, seeding medium was removed and replaced with fresh Greens. After a further 24 h, medium was removed from inside the inserts in order for the constructs to be at an aireliquid interface. Greens medium in the tissue culture wells was replaced every 24 h and constructs were used for experimentation after 14 days at aireliquid interface. 2.2.9. Infection of skin constructs Constructs were washed in antibiotic-free Greens medium for 72 h prior to infection. They were then burnt by the application of a heated metal rod, 4 mm in diameter, for 6 s immediately prior to infection in order to provide the bacteria with a mode of entry into the dermal tissue. 1  107 P.aeruginosa or S.aureus cells in 100 ml PBS per construct were pipetted into the inserts covering the epidermal surface. Infected constructs and non-infected controls were incubated in antibiotic-free Greens medium at 37  C/5% CO2 for the required time. 2.2.10. Incubation of AMMA-labelled polymer with infected skin After skin samples had been incubated with bacteria or PBS for 24 h, 100 ml of AMMA-labelled polymer (5 mg/ml) or PBS was pipetted directly onto the surface of the skin. Samples were incubated for 1 h at 37  C then imaged under UV light in a G:Box gel documentation system (Syngene) using Genesnap software (Syngene). 900 ml PBS was then added to each sample and the supernatants aspirated into fresh wells. Images were captured and processed on a Zeiss Axiovert 200 M inverted fluorescence microscope using Axiovision Rel 4.6 software. 2.2.11. Histology and viable bacterial counts Infected and non-infected constructs were bisected with sterile scalpels. Half of the sample was fixed in 10% formalin for >24 h then processed and paraffin embedded for histological analysis. The other half was weighed, then the tissue homogenised in 1 ml BHI broth. The resulting homogenates were diluted serially and used to perform viable counts of bacteria in the samples. Formalin fixed, paraffin embedded samples were sectioned to 6 mm thick on a microtome. Sections were stained with haematoxylin and eosin using standard techniques, and also Gram stained to visualise bacteria, again using standard techniques. 2.2.12. Bacterial attachment to hydrogel membranes Circular discs, 6 mm in diameter, of HB-PNIPAM-van hydrogel membrane, HB-PNIPAM-pmx hydrogel membrane or control HB-PNIPAM-COOH membrane were incubated with 100 ml S.aureus or P.aeruginosa at 1  1010 cfu/ml in PBS for 1 h at 37  C. Following incubation, membranes were washed in PBS to remove nonadherent bacteria and viewed under phase-contrast microscopy using a Zeiss Axiovert 200 M inverted microscope. Images were captured using Axiovision Rel 4.6 software and numbers of visible bacteria adhered to the membranes counted (n  5 images per membrane). 2.2.13. Live/dead staining of bacteria Membranes were incubated with 1 ml/ml Syto9 (Invitrogen) and 4 ml/ml propidium iodide for 20 min at room temperature, washed 4 times in PBS then viewed under fluorescence microscopy as above. 2.2.14. Membrane application to skin constructs Skin constructs were burnt as above and 1  107 cfu S.aureus or P.aeruginosa pipetted onto this wound, or PBS as control. Constructs were incubated for 45 min or 24 h at 37  C, then circular discs of HB-PNIPAM-van or epmx-B or eCOOH hydrogel membrane, 6 mm in diameter, were applied to the surface of the skin and incubated for 3 h at 37  C. Discs were removed and replaced with a fresh disc every hour. Bacteria adherent to the membranes were counted, viable bacteria remaining in the tissue sample quantified and tissue sections Gram stained as above. 3. Results 3.1. HB-PNIPAM-AMMA-pmx and HB-PNIPAM-AMMA-van solutions in contact with bacteria in an infected skin model Burn wounds in tissue engineered human skin were created. Uninfected controls were burned and incubated with PBS (a,b,c). Constructs were infected with P.aeruginosa (Fig. 1d, e, f) or S.aureus (Fig. 1g, h, i) as described previously [16]. After 24 h infection with bacteria and rinsing in PBS to remove unattached bacteria, 100 ml of polymer incorporating anthracene as a fluorescent marker (HB-PNIPAM-AMMA-pmx or HB-PNIPAM-AMMAvan) in solution (or PBS control) was pipetted directly onto the skin surface and incubated at 37  C for 1 h. Skin samples were then visualised under UV light. In samples which had been burned (Fig. 1a), infected with P.aeruginosa and incubated with HB-PNIPAM-AMMA-pmx, bright areas of AMMA fluorescence were seen in the areas of the burn injury (Fig. 1d), which microscopy showed to be the location of the highest concentration of bacteria. This was also the case for HB-PNIPAM-AMMAvan incubated with S.aureus infected skin (Fig. 1i). We also tested burned, uninfected samples to ensure that polymers were not simply attaching to burn injury-exposed extracellular matrix components such as fibronectin, which is known to be adhesive. In those samples there was no visible fluorescent area (Fig. 1b, c). Similarly, burned infected samples incubated with PBS rather than polymer showed no fluorescent areas (Fig. 1e, h). We then added a further 900 ml of PBS to each sample to aspirate polymer away from the skin. Examination of the resulting suspensions under a fluorescence microscope revealed that in eluates from infected samples clusters of fluorescent polymer were visible, but there were no clusters in eluates from uninfected skin (not shown). These findings, again indicate polymer collapse only in samples where the polymers had contacted the bacteria. We hypothesised that attachment of the polymer to bacteria in the infected skin model and its subsequent removal could physically debulk the bacterial load from the skin. Consequently, following infection and incubation with polymers or PBS as above, each sample was bisected and half examined histologically, while the other half was assessed for the bacterial viable count. In addition, the PBS wash supernatants were also serially diluted and counted. Histological analysis of the skin sections revealed that in infected samples incubated with PBS alone, a large number of bacteria were present particularly in the upper layers of the skin (Fig. 2a, e), in line with previous findings [16]. However, in samples that had been incubated with the appropriate polymer and the polymer then removed, there was a striking reduction in visible bacteria (Fig. 2b, f). Viable counts of bacteria present in the tissue samples confirmed this, showing a decreased number of bacteria present in those samples treated with polymer compared to those incubated with PBS alone (Fig. 2c, g). Also, the washed supernatants from the skin samples that had been incubated with polymer showed a higher number of bacteria than the supernatants from non-polymer treated tissue samples (Fig. 2d, h). Author's personal copy J. Shepherd et al. / Biomaterials 32 (2011) 258e267 261 Fig. 1. Anthracene-labelled polymer fluoresces on incubation with bacterial infected human tissue engineered skin (a) Gross view of burned skin construct within tissue culture insert. Arrow indicates injured area of epidermis. (b) and (c) Burned uninfected skin incubated with HB-PNIPAM-AMMA-pmx and HB-PNIPAM-AMMA-van respectively and viewed under UV light. (d), (e) and (f) Burned skin infected with P.aeruginosa and incubated with HB-PNIPAM-AMMA-pmx, PBS, and HB-PNIPAM-AMMA-van respectively. (g), (h) and (i) Burned skin infected with S.aureus and incubated with HB-PNIPAM-AMMA-pmx, PBS, and HB-PNIPAM-AMMA-van respectively. Arrow in (d) and (i) indicates bright spot which is the location of the burn injury. 3.2. Gram-positive and gram-negative bacteria binding to HB-PNIPAM-van or HB-PNIPAM-pmx hydrogel membranes Hydrogels were incubated with 100 ml S.aureus or P.aeruginosa at 1  1010 cfu/ml in PBS for 1 h at 37  C. Following incubation, membranes were washed in PBS to remove non-adherent bacteria and viewed under phase-contrast microscopy. Images were captured and numbers of visible bacteria adherent to the membranes counted (n  5 images per membrane). Visibly more S.aureus were bound to the HB-PNIPAM-van membrane as compared to control membrane, or P.aeruginosa bound to HB-PNIPAM-van membrane (Fig. 3aec). In contrast, P.aeruginosa bound preferentially to HB-PNIPAM-pmx membrane (Fig. 3f). Quantification of bacterial numbers demonstrated a significantly higher number of S.aureus bound to HB-PNIPAM-van membrane, and P.aeruginosa bound to HB-PNIPAM-pmx membrane, than any other test group (Fig. 3g). We then investigated whether bacteria bound to the hydrogel membranes were being killed or were merely being immobilised by binding to the membranes. To do this we performed live/dead staining over a 24 h time course, which differentiates between live and dead bacteria according to uptake of fluorochromes. Quantification of the percentage of live vs. dead bacteria revealed that the majority of bacteria were viable even after being adherent to either polymer for 24 h (Fig. 3h). Viable cells attached to membranes displayed little (P.aeruginosa) or no (S.aureus) growth over 24 h after incubation of the membranes at 37  C in optimal growth medium (Fig. 3i), and appeared to form a biofilm without a substantial release of bacteria into the surrounding medium, as indicated by the minimal increase in optical density of the growth medium compared to ‘free’ bacteria over the same time period (Fig. 3j). In contrast, previous studies with free antibiotic show that exposure to Vancomycin for 8 h would normally result in a 1e3 log decrease in viable bacteria, and at least a 2 log drop in numbers of P.aeruginosa treated with Polymyxin-B for 2 h [18e20] (i.e. >99% decrease). From the above findings we reasoned that physical binding of superficial bacteria in the wounded, infected skin model to the membrane might result in removal of bacteria when the membrane was removed. Consequently, after 45 min incubation at 37  C with 1  107 cfu S.aureus or P.aeruginosa to allow attachment to burn wounded models but not invasion of the tissue, circular discs of HB-PNIPAM-van or -pmx hydrogel membrane were applied to the Author's personal copy 262 J. Shepherd et al. / Biomaterials 32 (2011) 258e267 Fig. 2. Bacterial numbers are reduced in infected skin samples after incubation with HB-PNIPAM-AMMA-pmx or HB-PNIPAM-AMMA-van Gram stained sections of paraffin embedded burned skin were infected with P.aeruginosa, and incubated with either (a) PBS (control) or (b) HB-PNIPAM-AMMA-pmx polymer. Arrows in (a) indicate large numbers of Gram-negative bacteria present in upper layers of skin. Scale bar ¼ 200 mm. (c) Numbers of viable bacteria cultured from P.aeruginosa-infected skin tissue (n ¼ 7). (d) Numbers of viable bacteria cultured from supernatant of infected skin samples (n ¼ 5). Error bars ¼ SEM. (eef) Gram stained sections of paraffin embedded burned skin infected with S.aureus, and incubated with either (e) PBS (control) or (f) HB-PNIPAM-AMMA-van polymer. Arrows in (e) indicate large numbers of Gram-positive bacteria present in upper layers of skin. Scale bar ¼ 200 mm. (g) Numbers of viable bacteria cultured from S.aureus infected skin tissue (n ¼ 5). (h) Numbers of viable bacteria cultured from supernatant of infected skin samples (n ¼ 4). Error bars ¼ SEM. surface of the skin and incubated for 3 h at 37  C. The discs were then replaced hourly for 3 h. As shown in Fig. 4a numbers of S.aureus adherent to HB-PNIPAM-van membranes were significantly higher than those adherent to control membranes, or P.aeruginosa adherent to HB-PNIPAM-van membranes. Also, numbers of S.aureus adherent to HB-PNIPAM-van membranes increased with each application of membrane (Fig. 4a). Numbers of P.aeruginosa bound to HB-PNIPAM- pmx membranes were also higher than those bound to control membranes, but less bacteria were found bound with each application of fresh membrane (Fig. 4b). We suspect that this is due to the superior ability of P.aeruginosa to penetrate into the dermis out of reach of the membranes at the surface, thus P.aeruginosa was clearly visible within the dermis (Fig. 4e, f) Fig. 4cef show corresponding sections of skin models stained for bacteria. Author's personal copy J. Shepherd et al. / Biomaterials 32 (2011) 258e267 263 Fig. 3. Bacteria adhere to HB-PNIPAM-van and -pmx on a hydrogel membrane and remain viable over 24 h Phase-contrast microscopy images of (a) S.aureus adhered to HBPNIPAM-van hydrogel membrane and (b) control membrane, (c) HB-PNIPAM-pmx hydrogel membrane, and P.aeruginosa adhered to HB-PNIPAM-van (d), HB-PNIPAM-COOH (e) and HB-PNIPAM-pmx (f) hydrogel membranes. Numbers of adherent bacteria to each type of membrane are quantified in (g). Percentages of dead cells adhered to membrane at time points up to 24 h incubation are quantified in (h). Viable cells attached to membranes displayed a greatly reduced growth rate over 24 h (i), and were not released into the growth medium as evidenced by the negligible increase in optical density of the medium (j). Scale bars ¼ 20 mm. Error bars ¼ SEM. The number of viable bacteria remaining in the tissue samples following membrane treatment was also quantified (Fig. 4g). In this case, numbers of S.aureus were much lower in HB-PNIPAM-van membrane-treated samples compared to those treated with control membranes. Conversely, numbers in tissue samples treated with control membrane did not decrease with time. P.aeruginosa-infected skin treated with HB-PNIPAM-pmx membranes also contained less bacteria than control treated models, and the number of viable bacteria remained stable with each round of HB-PNIPAM-pmx membranes (Fig. 4h), whereas high numbers of bacteria were present at 3 h in the models treated with the control membranes. Again, we believe that this is due to the penetration of P.aeruginosa into the dermis. Gram stained paraffin sections of the infected skin samples showed less visible bacteria in skin treated with HB-PNIPAM-van membranes than those with control membranes (Fig. 4c,d). In order to observe the effects of the HB-PNIPAM-van membranes on skin constructs with a more established bacterial infection, skin constructs were burnt and then infected with 1  107 cfu S.aureus or P.aeruginosa as above. After 24 h incubation at 37  C to allow infection of the skin, circular discs of HB-PNIPAMvan or eCOOH control hydrogel membrane were applied to the surface of the skin as before and incubated for 3 h at 37  C. Discs were replaced hourly. Again, numbers of S.aureus adherent to HBPNIPAM-van membranes were significantly higher than those attached to control membranes or P.aeruginosa attached to HB-PNIPAM-van membranes (Fig. 5a). In this case however, the number of S.aureus adherent to HB-PNIPAM-van membrane decreased with each sequential application. We propose that this is due to the longer infection period (24 h vs. 45 min), which allows a higher proportion of the bacteria to penetrate into the dermis Author's personal copy 264 J. Shepherd et al. / Biomaterials 32 (2011) 258e267 Fig. 4. S.aureus and P.aeruginosa are reduced in infected skin after application of HB-NIPAM-van or epmx-B on a hydrogel membrane (a) Numbers of bacteria adherent to the HBPNIPAM-van membrane (n ¼ 5), (b) numbers of bacteria adherent to the HB-PNIPAM-pmx membrane (n ¼ 5), (c) Gram stained section of S.aureus infected tissue after two 1-h applications of HB-PNIPAM-van membrane, (d) Gram stained section of S.aureus infected tissue after two 1-h applications of HB-PNIPAM-COOH control membrane. (e) Gram stained tissue infected with P.aeruginosa after 3 hourly applications of HB-PNIPAM-COOH control or HB-PNIPAM-pmx membrane (f). (g) Viable S.aureus bacteria remaining in tissue following membrane treatment (n ¼ 5), (h) viable P.aeruginosa remaining in tissue following membrane treatment (n ¼ 5). Error bars ¼ SEM. Arrows indicate representative bacteria. Scale bars in c and d ¼ 200 mm, in e and f ¼ 50 mm. therefore out of reach of the membrane. In this case, over the 3 h treatment period the membranes remove surface bacteria available to them, the numbers of which decrease with each application. After a 45 min infection the numbers of adherent bacteria increase with each application since they are presumably multiplying at the skin surface at that early point. Viable counts of bacteria remaining in tissue following HBPNIPAM-van membrane treatment showed a decreased number of Author's personal copy J. Shepherd et al. / Biomaterials 32 (2011) 258e267 265 Fig. 5. S.aureus are reduced in infected skin after application of HB-PNIPAM-van on a hydrogel membrane following 24 h infection (a) Numbers of bacteria adherent to membranes (n ¼ 5), (b) numbers of viable bacteria remaining in tissue after application of membranes (n ¼ 5), c) and d) Gram stained tissue infected with S.aureus for 24 h after 3 hourly applications of HB-PNIPAM-van (c) or HB-PNIPAM-COOH control (d) membrane. Error bars ¼ SEM, scale bar ¼ 200 mm. S.aureus within the tissue at all time points compared to control membrane-treated samples (Fig. 5b). Rather than decreasing with each sequential application however, the numbers of S.aureus remained at a similar lowered level throughout the 3 h test period, as with the P.aeruginosa/HB-PNIPAM-pmx membrane 45 min incubation experiments. Again we surmise this is due to bacterial invasion of the dermis away from the membrane at the surface. Indeed, Gram staining of skin samples reveals that surface bacteria are removed in samples treated with HB-PNIPAM-van membrane compared with control membranes, but that some bacteria are found deeper within the dermis (Fig. 5c,d). 4. Discussion In this study, we describe the application of HB-PNIPAMs that are capable of selectively binding to either Gram-positive or Gramnegative bacteria in an in vitro infected human skin model. We demonstrate that they reduced the bacterial load in the model when added to the surface and subsequently removed some hours later. If these were incorporated in dressings they could be compatible with current wound care management practices. PNIPAM has an LCST of 32  C, whereas HB-PNIPAM-van has an LCST of above 60  C [3]. However, the LCST is reduced on binding to the target bacteria. By modifying the highly branched polymer with the addition of antibiotic end groups we have successfully altered the LCST so that binding of the appropriate bacteria at body temperature causes polymer collapse and consequent aggregation of bound bacteria. The key feature of these systems is that the binding-induced coil-to-globule transition converts the PNIPAM into a substrate to which bacteria are known to adhere. The bacteria are attached to the polymers by both the specific multiple binding interactions and the favourable physico-chemical environment provided by PNIPAM in the globule state. Bacteria are thereby ‘collected’ by binding to the polymer. This polymer could also be tethered onto diverse surfaces including wound dressings. There is no need to alter the temperature to induce these modified polymers in solution to collapse as the trigger is now the binding of the bacteria, which occurs over a temperature range compatible with wound management. Although we currently have no direct evidence for polymer collapse in hydrogel membranes, the modified polymers bound to solid hydrogel supports are also capable of binding bacteria in the infected wound model as discussed shortly. Although greatly reduced in number, viable bacteria were left within the tissue following application of polymer in solution, and we assume that in its current form the polymer can only debulk the wound bed at areas where it is in direct contact with bacteria, i.e. the upper surface layers. However, any reduction in bacterial load is encouraging. We found that skin samples incubated with soluble polymer had a significantly higher number of bacteria present within the washed supernatant than those incubated with PBS alone. We suggest that this is due to bacterial attachment to polymer, resulting in bacteria being washed away from the surface of the samples along with the polymer. Since simple washing-off of the polymer/bacteria complexes is sufficient to remove large numbers of bacteria from the skin, this supports the potential for the polymer to be applied as a clinically undemanding wound dressing. We hypothesised that repeated applications of the polymer to the same infected injury may result in an even more dramatic reduction in bacterial numbers. This would be the case in real-life situations, where wound dressings are changed regularly. Unless a patient is severely immunocompromised, the reduced numbers of bacteria remaining in the tissue could also be more effectively targeted by the patient’s own immune system than the larger numbers present pre-treatment with polymer. To simultaneously examine whether repeated applications of the polymer would indeed lead to reduced bacterial numbers within the skin samples, and to investigate the effectiveness of the polymer after attachment to a fixed surface, we created solid hydrogel membranes with either HB-PNIPAM-van, -pmx or control eCOOH attached. To examine the effect of hydrogel polymer on surface bacteria in a wound, after a short (45 min) incubation with S.aureus or P.aeruginosa Author's personal copy 266 J. Shepherd et al. / Biomaterials 32 (2011) 258e267 post-injury we found that numbers of S.aureus adherent to HBNIPAM-van membrane increased with each application of membrane, indicating enhanced bacterial removal with each subsequent round. This was supported by the decreasing numbers of viable bacteria remaining in the tissue after each application of HB-NIPAM-van. On the other hand, viable numbers of bacteria remaining in tissue after application of control membrane increased over the 3 h time period, indicating uncontrolled bacterial cell division over the course of the experiment. These data suggest that the hydrogel polymer is effective at removal of bacteria from the uppermost surfaces of wounds. In addition, specificity for Gram-positive organisms was confirmed as only low numbers of Gram-negative P.aeruginosa were adherent to the membrane. The opposite was true with HB-PNIPAM-pmx membrane, to which only low numbers of S.aureus adhered. Conversely, after a longer incubation with S.aureus post-injury, to simulate an established wound infection, the numbers of S.aureus adherent to HB-PNIPAM-van membrane decreased with each sequential application, presumably as each round of membrane application lowered the numbers of bacteria available near the surface to bind to the membrane. Numbers of S.aureus adherent to HB-PNIPAM-van membrane were significantly higher than those attached to control membrane, or P.aeruginosa attached to HB-PNIPAM-van membrane, only after 1 h since the total adhered diminished with each application to become closer to the control levels. The viable S.aureus remaining in the HB-PNIPAM-van treated tissue were reduced in number compared to control treated samples but remained fairly constant throughout. We assume that the membrane was able to remove the majority of the S.aureus from the surface of the wound but could not reach bacteria that had penetrated deeper into the tissue. These results mirrored those found for skin infected with P.aeruginosa for 45 min, which displays a greater capability for penetration into the dermis. Our previous findings with S.aureus infection in our skin model demonstrated that at 24 h post infection, the bulk of S.aureus remain located in the upper layers of the skin constructs, so we propose that particularly in the case of S.aureus infection, the greater part of the bacterial load could be removed by the polymer. Taken together, the data from the membrane experiments indicated that the polymer would be most efficacious if applied as rapidly as possible to the wound surface after injury, before bacteria are able to infect deeper tissue layers. Hydrogels have become increasingly used as wound dressings since their inception in the 1960s [21] due to their multiple desirable qualities, which include biocompatibility, relative low-cost, and ability to be applied in various forms such as gels or on gauze dressings [22,23]. In fact, after 24 h infection and 1 h application of polymer, HB-PNIPAM-van on hydrogel membranes was slightly more efficient at clearing viable bacteria from the tissue, leaving w500 cfu/mg compared to w1250 cfu/mg with the same polymer in solution (Fig. 5b c.f Fig. 3g). In previous published studies, other materials employing antibiotics linked to polymers have been developed [24e29]. In these materials ‘entire’ antibiotics are used and in some cases released, which could contribute to the development of antibiotic resistance. In our study Polymyxin-B was modified so that only the part of the molecule that binds bacteria is present, rendering it non-microbicidal. Data for HB-NIPAM-van indicate that even after 24 h exposure to the polymer, the majority of bacteria remain alive suggesting that the antibiotic cannot access sufficient target sites to kill bacteria presumably because it is tethered to the polymer. Also, this is likely to reduce markedly any potential contribution to selection of bacterial resistance and S.aureus growth on the hydrogel membrane was restricted to a slow-growing biofilm that did not appear to propagate into the growth medium. These are important considerations given the rise in antibiotic resistance over recent years and the natural propensity of bacteria to rapidly evolve resistance mechanisms. Ruiz et al. [30] recently described work in which Vancomycin was loaded onto a PNIPAM-modified polypropylene surface for use in medical implantation devices, in order to prevent biofilm formation. In this case, PNIPAM-grafted polypropylene films were soaked in Vancomycin solutions then dried. Under physiological conditions of temperature and pH, the PNIPAM mesh shrinks and releases the antibiotic. Consequently the former acts as a drug delivery device, similar to commercially available products such as wound dressings delivering topical polymyxin-B sulphate and bacitracin zinc to the wound, available without prescription in the United States of America [31]. These are in contrast to the current bacterial binding system that we have developed. 5. Conclusions In this study we demonstrate that stimulus-sensitive polymers can significantly deplete viable bacteria in a 3D tissue engineered model of infected human skin, by simple application and subsequent washing-off of the polymer. HB-NIPAM-van and -pmx can also be effectively immobilised onto the surface of biocompatible hydrogels, with no loss of the polymers’ wound debulking capabilities. Bacteria-sensitive polymers which have the capability to physically remove pathogens from infected wounds, without harming healthy tissue or contributing to selection of antibiotic resistance have enormous potential. Such polymers could be used to coat wound dressings, or be used for other topical applications, as well as for the removal of bacteria from liquid systems or surfaces. Acknowledgments We thank DSTL and EPSRC for funding. Appendix Figures with essential colour discrimination. Figs. 1 and 3 in this article may be difficult to interpret in black and white. The full colour images can be found in the on-line version, at doi:10.1016/j. biomaterials.2010.08.084. References [1] Klevens RM, Morrison MA, Nadle J, Petit S, Gershman K, Ray S, et al. Active bacterial core surveillance (ABCs) MRSA investigators. Invasive methicillinresistant Staphylococcus aureus infections in the United States. JAMA 2007;298:1763e71. [2] Burd A, Kwok CH, Hung SC, Chan HS, Gu H, Lam WK, et al. A comparative study of the cytotoxicity of silver-based dressings in monolayer cell, tissue explant, and animal models. Wound Repair Regen 2007;15:94e104. [3] Shepherd J, Sarker P, Swindells K, Douglas I, MacNeil S, Swanson L, et al. Binding bacteria to highly branched poly(N-isopropyl acrylamide) modified with vancomycin induces the coil-to-gobule transition. J Am Chem Soc 2010; 132:1736e7. [4] Kawaguchi H, Fujimoto K, Mizuhara Y. Hydrogel microspheres III. Temperature-dependent adsorption of proteins on poly-N-isopropylacrylamide hydrogel microspheres. Colloid Polym Sci 1992;270:53e7. [5] Zhang Y, Guan Y, Zhou S. Synthesis and volume phase transitions of glucosesensitive microgels. Biomacromolecules 2006;7:3196e201. [6] Carter S, Rimmer S, Rutkaite R, Swanson L, Fairclough JPA, Sturdy A, et al. Highly branched poly(N-isopropylacrylamide) for use in protein purification. Biomacromolecules 2006;7:1124e30. [7] Shen Z, Wei W, Zhao Y, Ma G, Dobashi T, Maki Y, et al. Thermosensitive polymer-conjugated albumin nanospheres as thermal targeting anti-cancer drug carrier. Eur J Pharm Sci 2008;35:271e82. [8] Zhou S, Chu B. Synthesis and volume phase transition of poly(methacrylic acid-co-N-isopropylacrylamide) microgel particles in water. J Phys Chem B 1998;102:1364e71. Author's personal copy J. Shepherd et al. / Biomaterials 32 (2011) 258e267 [9] de Gennes PG, Hervet H. Statistics of starburst polymers. J Physique Lett 1983;44:L351e60. [10] de Las Heras Alar con C, Pennadam S, Alexander C. Stimuli responsive polymers for biomedical applications. Chem Soc Rev 2005;34:276e85. [11] Stigter M, Bezemer J, de Groot K, Layrolle P. Incorporation of different antibiotics into carbonated hydroxyapatite coatings on titanium implants, release and antibiotic efficacy. J Control Release 2004;99:127e37. [12] Umeda M, Niwa M, Yamagami S, Kishimot T, Maekawa M, Sawada Y. Novel endotoxin adsorbing materials, polymyxin-sepharose and polyporous polyethylene membrane for removal of endotoxin from dialysis systems. Biomater Artif Cells Artif Organs 1990;18:491e7. [13] Blais BW, Yamazaki H. Use of polymyxin-coated polyester cloth in the enzyme immunoassay of Salmonella lipopolysaccharide antigens. Int J Food Microbiol 1990;11:195e204. [14] Cruz DN, Perazella MA, Bellomo R, de Cal M, Polanco N, Corradi V, et al. Effectiveness of polymyxin B-immobilized fiber column in sepsis: a systematic review. Crit Care 2007;11:R47. [15] Petsch D, Beeskow TC, Anspach FB, Deckwer WD. Membrane adsorbers for selective removal of bacterial endotoxin. J Chromatogr B Biomed Sci Appl 1997;693:79e91. [16] Shepherd J, Douglas I, Rimmer S, Swanson L, MacNeil S. Development of 3-dimensional tissue engineered models of bacterial infected human skin wounds. Tissue Eng Part C Methods 2009;15:475e84. [17] Rimmer S, Carter S, Rutkaite R, Haycock JW, Swanson L. Highly branched poly-(N-isopropylacrylamide)s with arginine-glycine-aspartic acid (RGD)- or COOH-chain ends that form sub-micron stimulus-responsive particles above the critical solution temperature. Soft Matter 2007;3:971e3. [18] Dixson S, Brumfitt W, Hamilton-Miller JMT. In vitro activity of combinations of antibiotics against Staphylococcus aureus resistant to gentamicin and methicillin. Infection 1984;13:35e8. [19] Flandrois JP, Fradel G, Carret G. Early stages of in vitro killing curve of LY146032 and vancomycin for Staphylococcus aureus. Antimicrobial Agents Chemother 1988;32:454e7. 267 [20] Tam VH, Schilling AM, Giao V, Kabbara S, Kwa AL, Wiederhold NP, et al. Pharmacodynamics of polymyxin B against Pseudomonas aeruginosa. Antimicrob Agents Chemother 2005;49:3624e30. [21] Wichterle O, Lim D. Hydrophilic gels for biological use. Nature 1960; 185:117e8. [22] Boateng JS, Matthews KH, Stevens HN, Eccleston GM. Wound healing dressings and drug delivery systems: a review. J Pharm Sci 2008;97:2892e923. [23] Eisenbud D, Hunter H, Kessler L, Zulkowski K. Hydrogel wound dressings: where do we stand in 2003? Ostomy Wound Manage 2003;49:52e7. [24] Klemm K. The use of antibiotic-containing bead chains in the treatment of chronic bone infections. Clin Microbiol Infect 2001;7:28e31. [25] Thakur RA, Florek CA, Kohn J, Michniak BB. Electrospun nanofibrous polymeric scaffold with targeted drug release profiles for potential application as wound dressing. Int J Pharm 2008;364:87e93. [26] Liu B, Hu J, Meng Q. Nonwoven supported temperature-sensitive poly (N-isopropylacrylamide)/polyurethane copolymer hydrogel with antibacterial activity. J Biomed Mater Res B Appl Biomater 2009;89:1e8. [27] Wang CC, Su CH, Chen CC. Water absorbing and antibacterial properties of N-isopropyl acrylamide grafted and collagen/chitosan immobilized polypropylene nonwoven fabric and its application on wound healing enhancement. J Biomed Mater Res A 2008;84:1006e17. [28] Changerath R, Nair PD, Mathew S, Nair CP. Poly(methyl methacrylate)-grafted chitosan microspheres for controlled release of ampicillin. J Biomed Mater Res B Appl Biomater 2009;89:65e76. [29] Krishnamurthy VM, Quinton LJ, Estroff LA, Metallo SJ, Isaacs JM, Mizgerd JP, et al. Promotion of opsonization by antibodies and phagocytosis of grampositive bacteria by a bifunctional polyacrylamide. Biomaterials 2006;27:3663e74. [30] Ruiz JC, Alvarez-Lorenzo C, Taboada P, Burillo G, Bucio E, De Prijck K, et al. Polypropylene grafted with smart polymers (PNIPAAm/PAAc) for loading and controlled release of vancomycin. Eur J Pharm Biopharm 2008;70:467e77. [31] Thornton Spann C, Taylor SC, Weinberg JM. Topical antimicrobial agents in dermatology. Clin Dermatol 2003;21:70e7.