Reusable nanoengineered surfaces for bacterial recruitment and decontamination

Materials exposed to nonsterile aqueous environments quickly become supports for attachment of microorganisms, which further elaborate into three-dimensional biofilms. The presence of biofilms themselves can have a direct detrimental effect on materials and devices, ranging from sensors, to implanted materials, to ship hulls; of equal concern, however, is the ability of biofilms to harbor and mediate the transfer of pathogens from a contaminated surface to humans and animals, and is of particular concern in healthcare,^1–3 the food supply and in engineered environmental systems. In addition, biofilms formed on such surfaces are able to readily transfer genes and are thought to play a particular role in acquisition and spread of multiple antimicrobial resistance genes in clinical settings.³ 

Cationic polymers and oligomers containing phenylene ethynylene (PE) backbones with pendant quaternary ammonium groups have been demonstrated to exhibit high levels of broad-spectrum antimicrobial activity.^4–6 The PE backbone closely mimics the structure of naturally occurring antimicrobial peptides found in most animals and, like these peptides, are thought to disorder the structure of the bacterial cellular membrane.^4–6 The activity of these compounds is dependent on both the structure of the oligo(phenylene ethynylene) (OPE) or poly(phenylene ethynylene) (PPE) and the lipid composition of the membranes with which they interact, resulting in low cytotoxicity for mammalian cells.^7,8 While most of the PPEs and OPEs, in analogy to antimicrobial proteins, are positively charged,^6,9 antimicrobial activity has also been observed with noncharged OPEs.¹⁰ While contact-mediated antimicrobial activity is considered an integral part of the biocidal action of PPEs and OPEs, enhanced light-activated killing, resulting from singlet oxygen production, is observed in some instances.^6–21 We have previously observed that cationic PPEs could attract large numbers of bacteria to a surface subsequently kill them with high efficiency,^13,15 and wished to exploit these capabilities in enhancing the ability to capture and kill bacteria.

Charged OPEs and PPEs are amenable to efficient assembly using layer-by-layer (LbL) techniques. LbL is a technique in which layers of oppositely charged polyelectrolytes are sequentially deposited on a surface.²² The resultant structure is stabilized by ionic interactions between the layers.²³ For these studies, we have selected to construct these layers from a positively charged imidazolium-functionalized poly(phenylene ethynylene) and a negatively charged poly(phenylene ethynylene) with proven light-activated biocidal activity as developed in our previous work.^13,19 As efficient as these compounds are against bacteria, cellular debris collects on the surfaces after the killing cycle.¹⁵ This debris results in reduced access of bacteria to the biocidal surface, but also serves as a possible conditioning layer for further bacterial attachment. Thus, combining these biocides with the ability to remove associated bacteria and debris is highly attractive.

Highly hydrated materials have been studied for nearly two decades for their ability to resist biofouling.^24–29 The association of water with such surfaces creates both steric and energetic barriers to the accumulation of cells on the surface.^27,28,30,31 In addition, sharp changes in the ability of polymers to form hydrogen bonds with water drives phase transitions in stimuli-responsive polymers (SRPs) that can result in release of attached bacteria and biofilms, resulting in tunable, efficient, fouling release strategies.^32–40 Recently, we have demonstrated that nanopatterned surfaces combining SRP and biocidal functionality are able to reversibly attach, kill, and release microbes.^41,42 These surfaces are composed of temperature-responsive polymer poly(N-isopropylacrylamide) (PNIPAAm) grafted from nanopatterned atom transfer radical polymerization (ATRP) initiators fabricated by UV-interferometric lithography, followed by introduction of biocides into the non-PNIPAAm voids. Bacteria are then attached and exposed to the integrated biocide and then released, along with cellular debris, by transitioning the PNIPAAm through its hydration transition temperature (T_t ∼ 32 °C).

We have previously demonstrated that surfaces that combine blends of light activated, uncharged OPEs and PNIPAAm prepared by vacuum deposition are able to effectively kill and release bacteria and bacterial debris.¹⁰ Herein, we explore whether segregation of the biocidal functionality of charged PPEs and the stimuli responsive functionality of PNIPAAm within well-defined nanoscopic areas results in more efficient killing and release. Within our experimental system, the LbL technique cleanly deposits a biocidal PPE layers in the valleys formed by surface tethered PNIPAAm brushes. Model Gram-negative and -positive bacteria are retained and killed upon exposure of the underlying LbL-deposited biocides and released upon transition through the transition temperature of the PNIPAAm. After the release step, the system can be reset to accumulate and kill further populations of bacteria, rendering it reusable.

N-isopropylacrylamide (NIPAAm), Cu(II)Br₂ (98% pure), 1,1,4,7,7-pentamethyldiethylenetriamine (PMDETA, 99% pure), and ascorbic acid (reagent grade, 20–200 mesh) were purchased from Sigma-Aldrich (St. Louis, MO). NIPAAm monomer was recrystallized twice from a benzene/hexane mixture and then dried under vacuum before use. The ATRP initiator (3-trimethoxysilyl) propyl 2-bromo 2-methylpropionate was purchased from Gelest (Morrisville, PA) and stored under dry condition until used. Silicon wafers and coverslips (size = 25 × 50 mm and thickness = 0.13 mm) were purchased from University Wafer and VWR, respectively. The biocidal PPEs PIM-2 and PPE-SO₃ (Fig. 1) were prepared and characterized as described in previous reports.^19,43 The samples were purified by aqueous dialysis followed by freeze-drying to obtain solid samples. Before use, the samples were redissolved in deionized water at 0.1 mM (concentration in polymer repeat units) (Fig. 1).

Escherichia coli K12 (ATCC 29425) and Staphylococcus epidermidis (ATCC 14990) were received as lyophilates from the American Type Culture Collection (Bethesda, MD), revived, and stored as frozen stock aliquots in Difco nutrient broth (NB) +20% glycerol at −80 °C. Experimental stock cultures were maintained on NB slants and were stored at 4 °C for up to 2 weeks. A single colony from the slants was incubated in 50 ml of NB and grown overnight with shaking at 37 °C. After growth, the bacterial culture was centrifuged at a relative centrifugal force of 11 952 × g for 10 min at 4 °C. The pellet was then suspended in 0.85% NaCl (for E. coli) or phosphate buffered saline (PBS) (for S. epidermidis). This washing procedure was repeated twice. The final concentrations of E. coli and S. epidermidis were ∼1 × 10⁸ cells/ml and ∼3 × 10⁷ cells/ml, respectively, as measured using a hemocytometer (C-chip CYTO Corp, Sunnyvale, CA) and phase contrast microscopy (Axioimager, Carl Zeiss Microimaging, Inc.) through a 40× objective.

Silicon wafers and cover slips were cleaned with “Piranha” solution [7:3(v/v) 98% H₂SO₄:30% H₂O₂; caution: piranha solution reacts violently with organic materials and should be handled carefully!] to remove the organic residue. The wafers were subsequently rinsed with an abundance of ultrapure water and dried under a dry nitrogen stream. The cleaned samples were immersed in 10 ml of anhydrous toluene containing the ATRP initiator-terminated silane (2 vol. %) at room temperature for 24 h. These surfaces were rinsed thoroughly with toluene and dried under a nitrogen flow.

Interferometric lithography (IL) was performed using a two-beam interference system (Lloyd's mirror setup) as reported previously.⁴⁴ Nanopatterns of ATRP initiator were fabricated by exposing ATRP initiator immobilized self-assembled monolayers (SAMs) to a diode-pumped, frequency-doubled neodymium-doped vanadate laser (Coherent, Verdi-V5) with a wavelength (λ) of 266 nm (energy dose of 13.9 J/cm²).

PNIPAAm polymer brushes were grafted from nanopatterned SAMs of ATRP initiators using activators regenerated by electron transfer (ARGET)-ATRP.^44,45 Briefly, samples were immersed into a solution containing 14 ml methanol, 14 ml H₂O, 2.5 g NIPAAm, 3.15 mg CuBr₂, 34.5 mg ascorbic acid, and 19.6 μl PMDETA for 6 min at room temperature. The samples were then removed from the solution, rinsed with an abundance of ultrapure water and methanol successively to remove both unreacted NIPAAm monomer and ungrafted PNIPAAm, and then dried under a nitrogen flow. As controls, PNIPAAm brushes were also grafted from unpatterned SAMs of ATRP initiators under identical polymerization conditions.

LbL deposition of PPE onto sample surfaces was conducted using the following protocol: (1) the sample surfaces were first incubated in water at 37 °C for 2 h to collapse the PNIPAAm; (2) the sample surfaces were submerged into a 0.1 mM aqueous PIM-2 solution (0.1 mM in water) for 10 min; (3) the sample surfaces were removed and immersed in water for 3 min followed by a second immersion in water for 1 min and a third immersion in water for 1 min; (4) the sample surfaces were then submerged into a PPE-SO₃ solution (0.1 mM in water) for 10 min; (5) the sample surface were removed and immersed in water in the same manner as described in (3) to achieve 1 layer of PPE composed of 1 PIM-2 layer and 1 PPE-SO₃ layer. The process of (2–5) was repeated for three times to achieve surfaces with three bilayers of PPE (Fig. 2). Finally, the surfaces were rinsed with an abundance of ultrapure water and dried under a nitrogen flow. Control surfaces were unpatterned PNIPAAm, unpatterned PIM-2/PPE-SO₃, and patterned PNIPAAm. Each type of surface (test surface and controls) was prepared on both silicon wafers and glass coverslips.

The elemental composition of surfaces was determined with a Kratos Analytical Axis Ultra x-ray photoelectron spectrometer (XPS) equipped with a monochromatic Al Kα source. High-resolution scans were acquired at a pass energy of 20 eV and a resolution of 0.1 eV. Survey scans were acquired with a pass energy of 160 eV and a resolution of 1.0 eV. All XPS data were analyzed using casa xps software. All binding energies were referenced to the main hydrocarbon peak designated as 285.0 eV. Peak resolution was performed using a linear peak base and symmetric 30/70 Gaussian–Lorentzian component peaks.

Ultraviolet–visible (UV–vis) absorption spectra were obtained using a Shimadzu UV-3600 spectrophotometer over a 350–500 nm range. A bare glass slide was placed in the reference chamber to remove the absorption background from the glass slide.

Tapping-mode topographical measurements of nanopatterned PNIPAAm surfaces in air before and after adsorption of lysozyme were obtained with a Digital Instruments Dimension 3100 atomic force microscope (AFM). Analysis (line profiling) was performed using nanoscope analysis software (Digital Instruments).

Contact angles were measured by a Rame-Hart model 100-00 contact angle goniometer either in air at 25 °C using sessile drop method or in water using the captive bubble method at 25 and 45 °C on a Rame Hart (Succasunna, NJ) goniometer with an environmental chamber and analyzed using drop analysis software. Contact angle values reported are the average of six replicates.

The thickness of films deposited on silicon wafers was measured with an M-88 spectroscopic ellipsometer (J. A. Woollam Co., Inc.). A three layer model was used for thickness estimation assuming a standard polymer (Cauchy layer) over SiO₂ and Si. The thickness values reported are the average of six replicates.

Attachment and detachment of bacteria were performed using established laboratory protocols^38–40 on the sample surfaces were assessed using E. coli suspension (1 × 10⁸ cells/ml in 0.85% NaCl) and S. epidermidis (3 × 10⁷ cells/ml in PBS). Briefly, prior to introduction of the samples surfaces, the cell suspensions were pre-equilibrated at 37 °C (E. coli) or 25 °C (S. epidermidis) in glass Petri dishes and sample coupons added (face up) to the bacterial suspensions where they were incubated for 2 h unstirred. After this time, coupons were rinsed gently with ultrapure water pre-equilibrated at 37 °C (E. coli) or 25 °C (S. epidermidis) to remove loosely attached cells and salts and then dried under a low-pressure stream of dry nitrogen. For bacterial detachment, the sample surfaces were washed under shear (estimated shear rate = 0.04 Pa)³⁹ with 60 ml 0.85% NaCl solution at 25 °C (for E. coli) or PBS at 37 °C (for S. epidermidis) delivered from a syringe, rinsed in ultrapure water, and dried. The attached bacteria were examined using a phase contrast optical microscope (Axioimager, Carl Zeiss Microimaging, Inc., Jena, Germany) through a 40× objective and images of ten randomly chosen fields of view were captured, and the density of attached bacteria was analyzed using imagej (http://imagej.nih.gov/ij/).^46,47 Three replicates were performed on each sample type and to obtain the average and standard deviation.

Light activated biocidal activity of the samples was assessed after incubation in the bacterial suspension, the films were then exposed to light (wavelength of 420 nm, 8 W) using UV lamp (Entela, Inc., UVGL-15) for 60 min at 37 °C (for E. coli) or at 25 °C (for S. epidermidis). Control experiments were performed in the dark at the same temperatures. The viability and surface morphology of attached bacteria was determined by live/dead staining assay and scanning electron microscopy (SEM), respectively.

A standard live/dead staining assay was performed to examine the biocidal activity of sample surfaces. Upon completion of the experimental treatments described above, the sample surfaces were immersed into a staining solution containing 1:1 mixture of 3.34 mM SYTO 9 (Invitrogen, Carlsbad, CA) and 20 mM propidium iodide (Invitrogen). After incubation at 37 °C (for E. coli) or at 25 °C (for S. epidermidis) for 15 min in the dark, the surfaces were rinsed with ultrapure water at 37 °C (for E. coli) or at 25 °C (for S. epidermidis) and examined by fluorescence microscopy (Axioimager) through a 40× air objective, and images of 15 randomly chosen fields of view were captured, and the relative numbers of live (green) versus dead (red) bacteria were analyzed using imagej.^46,47 For each sample, three replicates were performed to obtain the average and standard deviation.

The morphology of treated and untreated cells was determined using scanning electron microscopy after fixation and dehydration. The samples with attached bacteria were rinsed gently in ultrapure water to remove any loosely attached cells, fixed in 2.5% (v/v) aqueous glutaraldehyde for 2 h, dehydrated by ethanol (30%–100% v/v in water), and air-dried. Before characterization, the samples were sputter coated with a 5-nm layer of gold using a coated Denton Desk IV (Denton Vacuum, Moorestown, NJ) system. The surfaces were then examined using a Philips/FEI XL30 (Hillsboro, Oregon) field emission environmental scanning electron microscope at an accelerating voltage of 7 kV.

Solid surfaces are frequent culprits in the transmission of, and development of antibiotic resistance, in a number of diseases, ranging from hospital acquired infections^3,48 to foodborne illnesses^28,49 to those found in public places.^50,51 Over time, such surface-attached microbes can form tough, surface attached communities (biofilms), which are highly resistant to disinfection, increase the chances of genetic exchange (and thus the development of multiply resistant strains), and extend the lifetimes of those organisms on the surface.³ In addition, the development of multiply resistant pathogenic strains has outstripped the discovery of new antibiotics, and even the development of new discovery modalities has failed to result in new antibiotics for the treatment of Gram negative infections.^52,53 Given that the remaining effective antibiotics should be reserved for the most severe infections,⁵⁰ preventing transmission of pathogenic microbes is more important than ever and surface decontamination plays a key role in this effort.

PPEs are thought to disrupt the bacterial cell envelope, either by direct damage due to light-induced singlet oxygen,¹⁷ or by ionic disruption of the cell wall.^6,7,21 Because these types of damages are intrinsic to the basic structure of the cell envelope, resistance to these compounds is less likely to occur. Thus, these compounds offer an advantage over other biocidal surfaces.

Surface-attached biocides, including PPEs, can be effective at capturing and killing microbes,^11,13,15 but have the drawback of leaving dead cells and dead cell debris on the surface.¹⁵ The dead cells and debris not only block further cells from being attached and killed but also can serve as a locus for further colonization. Nanopatterned SRPs, such as PNIPAAm, have been shown to be effective at removing both live and dead bacteria from antimicrobial materials.⁴¹ Below, we discuss the results and significance of surface decontamination using PPE-PNIPAAm surfaces.

The samples used in the experiments discussed are described in Table I. Nanopatterned PNIPAAm supports were made using a combination of “top down,” interference lithography for patterning ATRP initiators followed by “bottom up” surface initiated polymerization strategies, as has been performed previously.^44,45 The following analyses suggest that we have produced nanopatterned surfaces which valleys between areas of PNIPAAm polymerization incorporated PPEs produced in an LbL assembly.

We monitored changes in molecular composition of the various samples described in Table I using XPS. Representative results of XPS survey spectra are shown in Table II. The degraded initiator surface (DI) shows similar chemical composition as that reported previously.⁴⁴ Significantly, within this sample, there is no nitrogen signal (predicted for both PNIPAAm and the PPEs) nor is there a sulfur signal (indicative of PPE-SO₃), but a significant silicon signal is present. When this surface is subject to LbL deposition of PPE (DI-PPE), the silicon signal is reduced. Ellipsometric measurements of the PPE bilayers suggest a thickness of 1.5 nm for three PPE bilayers (Fig. S2);⁵⁴ thus, the reduction in the silicon signal observed on the DI-PPE sample is reasonable. On the unpatterned PNIPAAm (UP), the silicon signal is completely absent, which is consistent with a uniform 13 nm-think overlay of PNIPAAm as we report below. On the nanopatterned PNIPAAm surface (NP), a reduction in silicon signal is also observed, due to the partial surface coverage of the degraded initiator, and finally on the NP-PPE surface, it is further attenuated. With the disappearance of the silicon signal, an increase in the nitrogen signal is observed on all samples. This signal is small on the DI-PPE, reflecting the relatively thin layer of PPEs on this sample, while larger on PPE-containing surfaces (NP, NP-PPE, and UP), reflecting the relatively thick layers the polymer on the surface. In addition, only those samples that contain PPE show the appearance of a sulfur signal (derived from PPE-SO₃) on DI-PPE and nanopatterned NP-PPE surfaces; it is not present on the surfaces that did not undergo LbL treatment (DI, NP, and UP). These results suggest that PPEs did not accumulate on the surface of PNIPAAm, and were restricted to the voids containing degraded initiator. These results are consistent with our previous results, which indicated that the degraded initiator within the grooves is hydrophilic and contains high levels of oxygen,⁴⁴ which are amenable to silanization,⁴¹ and thus are likely to contain ionizable silanol groups.

Supplementing the chemical information, AFM examination of the nanopatterned PNIPAAm (NP) and NP-PPE (Fig. 3) reveals that, although the period of the nanopatterns is largely preserved, the peak to valley height (PVD) is significantly decreased from 13.4 ± 0.9 to 11.1 ± 0.7 nm. This supports the hypothesis of accumulation of the PPE bilayers within the PNIPAAm channels. This change in depth is consistent with three PPE bilayers depositing using the estimation that a single bilayer of LbL-deposited polyelectrolytes varies from 0.3 to 0.7 nm depending on the material and deposition conditions.⁵⁵ Our measured ellipsometric thickness of three layers of PPE is 1.5 nm (Fig. S2); thus, the small, but statistically significant, difference in height is consistent with the deposition of PPE in the valleys. That the deposition of the PPEs was confined to the interstices between the PNIPAAm layer is confirmed by ellipsometric measurements of thickness, demonstrating that no increase in thickness was observed after performing LbL deposition procedures on unpatterned PNIPAAm grown by ARGET-ATRP on silicon.

The UV–visible spectra of NP-PPE and control surfaces are shown in Fig. 4. No significant absorption of 410 nm light was observed for LbL-deposited PPE on unpatterned PNIPAAm surfaces (UP-PPE), whereas that deposited on degraded initiator (DI-PPE) showed strong characteristic absorbance at this wavelength. NP-PPE also showed an absorbance at this wavelength, although less intense, consistent with the reduced amount of available degraded initiator on the nanopatterned samples. Upon addition of each PPE bilayer 410 nm absorbance increased on DI-PPE, but not on unpatterned PNIPAAm controls (Fig. S1). These changes were in accordance with previous observations that a 0.005 AU change in the spectrum is observed for each bilayer of PPE added.⁵⁶ The 0.015 AU peak for the DI-PPE is, thus, consistent with a fully formed three bilayer structure. That on the NP-PPE is decreased due the interspersed/PNIPAAm nanopatterns. These results confirm that the LbL deposition of PPE is likely confined to the degraded initiator regions between the nanopatterns of PNIPAAm. These data are corroborated by ellipsometric methods, in which no increase in thickness was observed on UP-PPE surfaces, whereas a step change between each successive round of deposition was seen on the DI-PPE surfaces (Fig. S2). Examination of the UV spectra after water immersion indicates that PPE is relatively stable on NP surfaces in water either above or below the lower critical solution temperature (LCST; ∼32 °C) of PNIPAAm (Fig. S3).

An important functional feature of surface grafted PNIPAAm is a change in wettability upon transition through its LCST (∼32 °C) resulting from changes in the hydrogen bonding in the polymer changes from water below the LCST to intrachain interactions above.^57,58 This property is preserved for NP-PPE (Fig. 5). Measurements of the contact angle of a captive air bubble (θ_CB) under water in and the temperature changed in situ. As expected, no significant change was observed on the DI-PPE samples (θ_CB ∼ 52°); NP-PPE samples exhibited a change in θ_CB from 32° to 40°; a slightly smaller change in θ_CB was measured for the NP brushes alone, from 28° to 40°, indicative of a hydration transition within the nanopatterned PNIPAAm brush.

Bacterial attachment onto PNIPAAm surfaces seems to be at least partially dependent on the wettability of the polymer surface, and subsequent release depends on a change in wettability to one less favorable for initial attachment upon transitioning through the LCST.^39,59 Because we observed such a substantial change in θ_CB, we expected both the NP and NP-PPE to attach and release bacteria. Attachment of E. coli to NP-PPE (2293 ± 479 cells/mm² over 2 h) was slightly higher as compared to NP surfaces (1147 ± 414 cells/mm²) [Fig. 6(a)]. This result is consistent with previous observations that LbL-PPE surfaces are highly attractive to bacteria.¹³ Significantly more bacteria attached to the DI-PPE surfaces than to the NP surfaces, which correlates well with previous observation⁴¹ in which nanopatterns themselves confer partial antifouling properties. These surfaces also attached comparatively more E. coli, when adjusted for initial concentration in the suspension than did similarly constructed surfaces using a random distribution of PNIPAAm and CPE, when adjusted for initial bacterial concentration.¹⁰ SEM examination revealed that distribution of these rod-shaped cells was random over the nanopatterned surface (Fig. 8) with no preference for alignment along the grooves, contrary to what is often observed with microstructures.⁶⁰ 

Upon transition below the LCST, no statistically significant bacterial release was found on DI-PPE surfaces [Fig. 6(b)]. In contrast, PNIPAAm containing surfaces (NP-PPE and NP) released 66 ± 3% and 81 ± 9%, respectively, similar to what we have previously observed for nanopatterned NP-biocide surfaces.^41,44 These results are also consistent with release from randomly distributed OPE-PNIPAAm surfaces.¹⁰ 

We have previously demonstrated that PPE-based polyelectrolytes can exhibit both dark and light bacteriocidal activity,^11–13 and that these properties are preserved when PPEs are immobilized on a surface.^11,15 We demonstrate here that LbL assembly of PPEs onto nanopatterned PNIPAAm results in materials with biocidal activity.

NP-PPE and DI-PPE exhibited the similar levels of dark and light killing (Fig. 7), as assessed by a standard live-dead fluorescent cell permeability assay. PPEs are known to exhibit two types of biocidal activity depending on the light regime to which they are exposed.⁷ While light-induced killing is thought to be the result of generalized destruction caused by production of singlet oxygen,⁷ dark activity is thought to be caused by direct disruption of the protein structure within the cellular envelope of Gram negative organisms, specifically E. coli.²¹ 

The percent killed is somewhat reduced from PPEs in solution, which can be greater than 99%;¹⁴ however, it is similar to that seen for quaternary ammonium compound/PNIPAAm surfaces (73%).⁴¹ This killing ratio is also similar to light-induced biocidal activity of unpatterned OPE-PNIPAAm surfaces that are optimized for both killing and release.¹⁰ NP surfaces containing only PNIPAAm exhibited ∼15% killing of E. coli, which cannot eliminate a slight cytotoxicity on this surface, as the typical percentage of dead cells in an overnight culture for this strain is ∼5%.¹⁴ SEM examination of cells on all surfaces revealed that the cellular envelopes of bacteria attached to PPE containing surfaces (DI-PPE and NP-PPE) were degraded; those attached to the NP surfaces are seemingly intact (Fig. 8). These results are consistent with dark-induced damage of E. coli by cationic PPEs.⁷ 

One major difference between this and previous studies is that the top layer of the bilayers deposited here is an anionic PPE (PPE-SO₃), whereas most previous studies have emphasized cationic PPEs.^11–18,21 To that end, we examined the biocidal activity of a 2.5 bilayer surface, in which the top layer was cationic PIM-2 (Fig. S4); there was no significant improvement of killing with the cationic PPE as the top layer. Cationic PPEs and OPEs are thought to exert their biocidal activity by clustering negatively charged, negative curvature lipid membranes, thus disrupting the structure or, by an ionic exchange process destabilizing the cell wall structure.⁹ The latter activity may be also operative in the three bilayer system, with anionic moieties exposed and the lack of enhanced biocidal activity when PIM-2 is exposed may suggest that it is also the major mechanism for the 2.5 bilayer system as well.

Gram positive and Gram negative cells have different cell envelope structures, and as a result, CPEs exert different show different levels and possibly mechanisms of biocidal activity against these organisms.^7,14–16 We, therefore, tested the efficacy of NP-PPEs against a model Gram-positive organism, S. epidermidis. While attachment of S. epidermidis was similar to that for E. coli (Fig. S5), detachment of S. epidermidis approached 80% on PNIPAAm-containing surfaces.

Biocidal activity of DI-PPE and NP-PPE surfaces was also observed against S. epidermidis. DI-PPE and NP-PPE killed attached bacteria (Fig. S6). Intrinsic killing by PNIPAAm on the NP surfaces was also about 15%. There was neither a significant difference between light and dark killing of NP-PPE or DI-PPE against these organisms nor was there a difference in killing efficiency over E. coli. Once again, damage to the cell was observed on SEM micrographs (Fig. S7); however, the damage was much less obvious on the NP-PPE surfaces than on the DI-PPE surfaces. We propose that this is due to the decreased contact of the cell surface of the spherical S. epidermidis cells with the PPE part of the NP-PPE surface {at most 1 stripe of PPE [Fig. S4(c)] compared to four for a typical E. coli (Fig. 8) features}.

The greatest promise for nanopatterned, biocide-containing surfaces is in their ability not only to attach and kill cells, but also to release the dead cells and debris afterward, thus refreshing the biocidal surface. We, therefore, tested the ability of these surfaces to undergo repeated rounds of attachment, killing, and release and examined whether efficiency of these processes was affected. Figure 9 compares killing and release of E. coli on NP-PPE over three repetitive cycles. There is a small, but statistically significant, decrease in both efficiency of killing (from 65 ± % in cycle one to 56 ± 3% in cycle three) and release (66 ± 3% in cycle one to 52 ± 4% in cycle three). Each successive cycle of attachment resulted in ∼2000 new cells attaching, indicating that the surface competency for attachment did not decrease over these cycles. Accumulation over time of attached cells from previous cycles resulted in a greater number of cells at the final cycle.

Taken together, these results demonstrate that nanopatterning of PNIPAAm brushes followed by layer-by-layer deposition of PPE films results in surfaces in which the PPE segregates into the non-PNIPAAm containing voids. The resulting structures exhibited similar biocidal and release properties when compared with similar coatings with smaller, biocidal molecules. This result suggests a facile method for engineering surfaces that attach, kill, and release bacteria. The similar killing efficiency when either anionic or cationic PPEs were exposed suggests that the biocidal mechanism proceeds through disruption of the cell envelope of the bacteria.

This work was supported by funding from the Defense Threat Reduction Agency (Grant No. HDTRA-1-11-1-0004), the Office of Naval Research (N00014-13-1-0828), and from the NSF's Research Triangle MRSEC (DMR-1121107).