DOI: 10.1002/ (adhm201600948) Article type: Full Paper
Bacterial Interactions with Immobilized Liquid Layers
Yevgen Kovalenko, Irini Sotiri, Jaakko V. I. Timonen, Jonathan C. Overton, Gareth Holmes, Joanna Aizenberg, Caitlin Howell*
Y. Kovalenko, I. Sotiri, G. Holmes Wyss Institute for Biologically Inspired Engineering, 60 Oxford Street, Cambridge, MA, 02138 USA Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, 9 Oxford Street, Cambridge, MA, 021383 USA
Prof. J. V. I. Timonen Department of Applied Physics, Aalto University, Puumiehenkuja 2, 02150, Espoo, Finland Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, 9 Oxford Street, Cambridge, MA, 021383 USA
J. C. Overton Department of Chemical and Biological Engineering, University of Maine, 5737 Jenness Hall, Orono, ME, 04469 USA
Prof. J. Aizenberg Wyss Institute for Biologically Inspired Engineering, 60 Oxford Street, Cambridge, MA, 02138 USA Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, 9 Oxford Street, Cambridge, MA, 021383 USA Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, MA 02138 Kavli Institute for Bionano Science and Technology, Harvard University, 12 Oxford Street, Cambridge, Massachusetts 02138 USA
Prof. C. Howell Department of Chemical and Biological Engineering, University of Maine, 5737 Jenness Hall, Orono, ME, 04469 USA Graduate School of Biomedical Science and Engineering, University of Maine, 42 Stodder Hall, Orono, ME, 04469 USA Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, 9 Oxford Street, Cambridge, MA, 021383 USA E-mail: caitlin.howell@maine.edu
Keywords: Oil-infused polymers, biofilm, bacteria, infection, SLIPS
Bacterial interactions with surfaces are at the heart of many infection-related problems in
healthcare. In this work, the interactions of clinically-relevant bacteria with immobilized
1

liquid (IL) layers on oil-infused polymers are investigated. Although oil-infused polymers reduce bacterial adhesion in all cases, complex interactions of the bacteria and liquid layer under orbital flow conditions are uncovered. The number of adherent Escherichia coli cells over multiple removal cycles increases in flow compared to static growth conditions, likely due to a disruption of the liquid layer continuity. Surprisingly, however, biofilm formation appears to remain low regardless of growth conditions. No incorporation of the bacteria into the layer is observed. Bacterial type is also found to affect the number of adherent cells, with more E. coli remaining attached under dynamic orbital flow than Staphylococcus aureus, Pseudomonas aeruginosa under identical conditions. Tests with mutant E. coli lacking flagella confirm that flagella play an important role in adhesion to these surfaces. The results presented here shed new light on the interaction of bacteria with IL layers, highlighting the fundamental differences between oil-infused and traditional solid interfaces, as well as providing important information for their eventual translation into materials that reduce bacterial adhesion in medical applications.
1. Introduction Infections contracted in clinical settings are often a consequence of bacterial adhesion and biofilm formation on implants, extracorporeal devices, and surgical tools. Treatment and prevention of such fouling requires the administration of drugs that contribute to the everincreasing problem of antibiotic resistance,[1] and can also increase risk of patient mortality.[2] Slippery liquid-infused porous surfaces (SLIPS), inspired by the carnivorous Nepenthes pitcher plant, were recently introduced as a new approach to prevent bacterial biofouling in both industrial and medical applications without the need for administering antibiotics.[3–5] This approach represents a paradigm shift in materials science, where instead of a solid surface, bacteria are presented with an inherently dynamic and self-healing immobilized liquid (IL) layer that is held in place by a chemically matched substrate.[3,6,7]
2

Epstein et al. were the first to demonstrate that the presence of an IL layer of

fluorinated oil on ePTFE significantly reduced biofilm coverage of S. aureus and E. coli (by

97.2% and 96%, respectively) after 48 hours of growth under static conditions, compared to

ePTFE without an IL layer.[4] Similarly, P. aeruginosa biofilm coverage was reduced by

99.6% after 7 days of growth under gentle flow (10 ml min-1).[4] Other works have reported

similar results for bacterial biofilm formation on IL layers on oil-infused polymers: S. aureus

under static conditions[8] and P. aeruginosa under flow.[9] Recently, Li et al. studied various

strains of P. aeruginosa grown on porous liquid-infused surfaces under flow in a continuous

reactor, finding that most resulted in less than 1% biofilm coverage.[10] However, they also

discovered that the strain P. aeruginosa with the highest drug resistance also showed the

highest surface coverage, 12% after 7 days of incubation, pointing to potential differences

among different bacterial types.

Bacterial adhesion and biofilm formation on solid surfaces is a complex

phenomenon.[11,12] Generally, bacterial adhesion can be classified as either reversible or

irreversible, as judged by the ability to remove surface-associated bacteria with a gentle

rinse.[11] Reversible attachment can be mediated by chemiosmotic forces and cellular surface

proteins, while irreversible attachment is achieved through production of extracellular

polysaccharides and ligand-specific protein interactions.[13] For flagellated species such as E.

coli and P. aeruginosa, flagella can play an additional role in both reversible and irreversible

attachment on solid surfaces.[14–16] In nearly all cases, however, bacteria that interact with a

solid will either irreversibly attach to a surface and begin producing extracellular polymeric

substances (EPS) to eventually form a biofilm, or they will leave the surface and return to the

planktonic state.[17] The shear stress created by flow conditions has been shown to enhance

bacterial cell attachment to surfaces for a number of different species in a variety of ways,

including the activation of catch-bonds in E. coli[18] and increasing the surface residence time

of P. aeruginosa.[19]

3

Little detail is currently known about how bacteria interact with IL layers on oilinfused substrates, especially under complex flow conditions. However, this information is necessary to enable the rational design and optimization of materials that take advantage of the unique properties of IL layers in medical applications. In this work, we investigate in detail bacterial interactions with IL layers, comparing adhesion under both static and complex flow (dynamic) conditions, differences across three clinically-relevant species, and, for E. coli, the effect of flagella. We focus specifically on IL layers created using silicone oil-infused polydimethylsiloxane (PDMS), as this is an easily fabricated material made from non-toxic, widely available components already used in medicine. Furthermore, various types of this material have already proven useful as fouling-release coatings in marine settings. [20,21] However, in this work infused PDMS was used specifically to serve as an anchor for an immobilized liquid overlayer.
Using iterative sonication-vortexing removal cycles to compare numbers of adherent cells, we show that while significantly fewer E. coli remain attached to the surface when grown under static conditions, this difference intially disappears in cultures grown under dynamic conditions, with surfaces showing similar number of adherent cells as solid controls. Interestingly, no equivalent increase in biofilm formation is observed. Visualization of the surfaces suggests that the increase in adherent cells is due to a disruption of the liquid layer. However, further removal cycles show that these cells are poorly attached and more easily removed from these surfaces than from controls. We further show that E. coli appears unique in its ability to adhere to IL layers under dynamic conditons, as tests on S. aureus and P. aeruginosa show decreased numbers of adherent cells under similar dynamic conditions. Further work with flagella-deficient E. coli mutants suggests that these differences may be partly due to flagella. The results provide important information about the factors affecting the interaction of bacteria with IL layers, and will help inform their future development and application as anti-adhesion surfaces in healthcare settings.
4

2. Results and Discussion 2.1. Dynamic versus Static Growth Conditions.
Figure 1 shows comparisons between the regular PDMS and PDMS infused with silicone oil after exposure to a clinical isolate of E. coli for 48 h under static or dynamic conditions. Dynamic conditions were created by placing the samples on an orbital shaker at 100rpm. In Figure 1A, representative images of the surfaces with CV-stained biofilms are shown. Although both non-infused surfaces show biofilm, the dyed biomass on the sample grown under static conditions is darker and more pronounced than that of the sample grown under dynamic conditions. This is likely due to the fact that more bacteria can settle on the surface in the absence of any flow to remove them.
Figure 1. E. coli biofilms grown under static (upper row) and continuous orbital shaking (lower row) conditions: (A) Representative images of the surfaces with the biofilm stained purple. Scale bars are 0.5 cm. (B) Percent biofilm coverage from image analysis. (C) Colony-forming unit counts.
5

The infused PDMS surfaces exposed to both static and dynamic conditions appeared cleaner than their respective non-infused controls. However, there does appear to be a small amount of biofilm present on the infused surface exposed to dynamic conditions, while the infused surface exposed to static conditions appears completely clean, as previously reported.[4,8,9] The biofilm coverage results are quantified in Figure 1B, which confirms the somewhat greater amount of biofilm coverage for non-infused samples exposed to static conditions compared to non-infused samples exposed to dynamic conditions (96.4 (±4.7)% versus 51.0 (±24.2)%, respectively, P = 0.046). However, this difference is slight compared to the drastic difference in surface coverage between infused versus non-infused samples: coverage drops to 0.3 (±0.1)% for infused samples under static conditions and to 4.5 (±2.6)% for infused samples under dynamic conditions (P < 0.05).
Figure 1C shows the colony-forming unit (CFU) counts for all of the samples after dipping to remove planktonic bacteria, followed by one vortex (30 s) and sonication (3 min) cycle to remove adherent bacteria. The results are normalized such that the average of the non-infused PDMS samples is equal to 1, to facilitate a more accurate comparison between the non-infused and infused samples. Although the infused samples exposed to static conditions show a decrease of approximately two orders of magnitude under this treatment, the samples grown under dynamic conditions show no decrease in adherent bacteria with infusion. While the presence of a small amount of biofilm coverage on the infused dynamic samples may explain the presence of a small number of adherent bacteria, it is surprising to see the number of adherent bacteria on these samples is similar to non-infused control samples, where the percent biofilm coverage is 10 times larger. These results suggest that either (1) the E. coli are adhering to the IL layer under dynamic conditions, but not producing EPS as they are on the controls, (2) the EPS that is produced is removed while the producing bacteria remain behind, or (3) the bacteria and EPS are becoming mixed into the oil layer.
6

To further investigate the lack of difference between CFUs measured on infused versus non-infused samples exposed to dynamic conditions, we visualized the samples using confocal microscopy and a triple-reporter system. In this system, the PDMS fluoresced green, the silicone oil overlayer orange, and the E. coli red. The results are shown in Figure 2. In these images, the red bacteria appear elongated due to the weak flow of the lubricant and aqueous phase during imaging that typically takes several tens of seconds (z-stack). These flows originate from non-complete equilibration of the samples before imaging. The apparent motion of the bacteria is thus not due to their intrinsic swimming, but rather due to their motion with the flow of the surrounding media, and thus all bacteria seem to move uniformly in the images. The samples that were exposed to static growth conditions showed a layer of silicone oil covering the PDMS substrate, regardless of where on the sample the image was taken (Figure 2A, Figure S1). However, the layer was not necessarily uniform in thickness, ranging from approximately 15 to 35 µm. The samples kept under dynamic conditions showed even more variable thicknesses, however, with localized dewetting causing the oil to pool (Figure 2A, Figure S1). This pooling was only observed when the samples were placed under aqueous medium. In air, the oil layers self-healed as expected. Nevertheless, such pooling is in agreement with the increase in biofim coverage observed for infused samples exposed to dynamic conditions (vs static conditions) shown in Figure 1 and 2, as bacteria may
7

be able to adhere more strongly to the areas where the IL layer is depleted.
Figure 2. Confocal images of the liquid overlayer. (A) Greater differences in thickness appear within a sample after 48 h incubation with bacteria under dynamic conditions compared to static conditions. Scale bars are 5 µm. (B) Top and side views of an infused surface after exposure to E. coli for 48 h under either dynamic or static conditions. More bacteria (red) appear on the dynamic sample versus the static sample, however the bacteria are always localized on the surface of the oil layer rather than incorporated into it. Scale bars are 20 µm.
Overall, the visualization results suggest that while static conditions yield a more consistent oil overlayer than dynamic conditions, the system is still sensitive to various conditions and may not be perfectly uniform. The difference in the overlayer observed for the samples kept under dynamic conditions may have been due in part to the presence of turbulent flow where the fluid came in contact with the edges of the square samples, which would have been more pronounced near the outer edges of the dish due to the higher velocity of fluid flow at those points. Previous work visualizing the stability of immobilized fluorinated liquid
8

layers under laminar flow in water showed that they remain intact under these conditions, and did not appear to be completely stripped away even at shear stresses of up to 2.32 kg/(m·s2).[22] However, the fact that the layers examined in that work were not exposed to turbulent flow conditions and were only investigated in contact with pure water, not complex medium, may explain the differences. The use of silicone oil in the work presented here, as opposed to the more inert fluorinated oils used in other studies, may have also contributed. Recent results by other researchers using silicone oil layers has suggested that a disruption of the layer at edges and corners can result in increased bacterial presece in these areas.[23]
Visualization of bacterial attachment on samples exposed to dynamic conditions showed individual bacterial cells to be present only at the surface of the oil layer; no bacteria were observed to be incorporated into the oil (Figure 2B, side view; Figure S2). However, more bacteria were present on the dynamic samples compared to the static samples (Figure 2B, top view), in agreement with the increase in CV absorbance observed in Figure 1B. These results support the hypothesis that E. coli are adhering more strongly to the infused surface under dynamic conditions, rather than becoming incorporated into the oil layer. However, whether or not they are producing EPS that is then separated from the surface, or whether their EPS production is inhibited, are questions that should be the subject of future investigations, and will likely require in-depth analysis of the activation of relevant gene pathways, such as Cpx.[24] Nevertheless, it is known that the restriction of flagellar motion by a solid surface can serve as a signal for some bacteria to form a biofilm.[25,26] One explanation, therefore, may be that the presence of a liquid layer, even a partial one, continues to allow flagellar motion to a limited degree and does not trigger the full cascade of biofilm-forming signals within the E. coli cells.

2.2. Bacterial Adhesion

9

To give a more complete picture of the attachment of E. coli to the infused PDMS surfaces relative to non-infused controls, we tested their relative adhesion strength through multiple vortexing and sonication cycles followed by CFU counts. The combination of sonication and vortexing was designed to remove the majority of bacteria in one cycle,[27] implying that the remaining bacteria would be firmly attached to the surface. Confocal microscopy of the surfaces after the initial gentle rinse to remove planktonic organisms and again after the first removal cycle showed that the IL layer was completely removed by this vortex-sonication process, at least to the level visible by microscopy (Figure S3).
The absolute CFU counts for both infused and non-infused PDMS samples under dynamic and static conditions are shown in Figure 3A, while the results normalized with respect to non-infused controls are given in Figure 3B. The normalized results were found to fit well to a modified stretched exponential function. The stretched exponential function was originally developed to describe relaxation processes in complex systems such as the adsorption and desorption of polymers on a surface,[28] but has also recently been used to describe the desorption of bacteria from a surface under flow.[19] This equation relates material or cells being adsorbed or desorbed from a surface to time. In our case, the modified function is given by

������������������$%&' 	 = (1 − ������) ∗ ������1$2	

(1)

where ������ represents the change in CFU counts compared to the non-infused control upon the first removal cycle (with a more positive value indicating a larger initial decrease), ������ the cycle number, and ������ is a shape parameter which represents the severity of the drop in CFU counts
in subsequent cycles (with a larger value indicating easier subsequent removal). Fitting with this function was intended to further parameterize the normalized removal curves and give a more quantitative basis for comparison. All resulting values are given in Table S1.
10

Figure 3. CFU Counts of bacteria removed from PDMS and infused PDMS via vortexsonication cycles (removal cycles) after 48 h of growth under either static or dynamic conditions. (A) Absolute CFU counts (B) CFU counts from infused PDMS normalized to non-infused PDMS and fit with an exponential decay curve. The dotted line gives the values for non-infused PDMS at each removal cycle.
For samples grown under static conditions (Figure 3, static), the results reveal a
decrease of approximately two orders of magnitude upon the first vortex-sonication cycle (������ =
0.96, corresponding to a 98% decrease in removed CFUs), and a drop to zero CFUs upon the
third removal cycle. These results support the hypothesis that when grown under static
conditions, the few bacteria that remain on a liquid-infused surface after removal from the
culture medium are poorly attached and can be relatively easily stripped off compared to non-
infused controls.[4,8,9]
In contrast, bacteria attached to the infused surfaces under dynamic conditions (Figure
3, dynamic) appeared to show greater persistence with multiple cycles of vortexing and
sonication. Unlike samples grown under static conditions, there was no decrease in CFUs 11

removed from these samples compared to their non-infused controls upon the first removal cycle (������ = -3.47, P = 0.203). However, there was a significant difference at the 4th and final removal stage (������ = 0.73, P = 0.017), suggesting that E. coli that were adherent on the infused surfaces were still ultimately more easily stripped off and therefore less adherent compared to those attached to non-infused control surfaces.
Softening of the PDMS substrate by silicone oil swelling may be one possible explanation for the easier removal of bacteria incubated even under dynamic conditions. However, a recent study of E. coli attachment on PDMS substrates of varying stiffness showed that E. coli appeared to adhere more strongly to softer surfaces rather than harder surfaces.[29] Therefore, it is more likely that the reduced adhesion strength observed for E. coli is due to the presence of the infusing silicone oil and its transport to the interface rather than the softening of the PDMS matrix as a result of swelling. 2.3. Species-specific assays
As previously mentioned, different bacterial strains use different methods to attach and anchor themselves to surfaces. To test for differences in bacterial adhesion to liquid-infused surfaces, samples were exposed to the gram-positive coccal bacterium S. aureus and the gram-negative rod P. aeruginosa under dynamic conditions. One of the strengths of liquidinfused surfaces is the fact that this primarily physical approach to resisting bacterial attachment has been previously shown to be fairly consistent across a range of species.[4] However, most reports of bacterial attachment to liquid interfaces have tended to focus on a single species.[8–10] Furthermore, of these, only P. aeruginosa has been thoroughly tested under continuous flow conditions.[4,9,10]
The biofilm coverage results and normalized removal curves for S. aureus and P. aeruginosa grown under dynamic conditions are shown in Figure 4. Figure 4A shows the overall CV-stained biofilm coverage after only a dip in water to remove planktonic bacteria. S. aureus and P. aeruginosa are known to form thick, persistent biofilms, and the coverage
12

results show a clearly thicker biofilm present on non-infused samples exposed to these two organisms compared to the E. coli biofilms in Figure 1. However, for both species, biofilm formation is significantly lower on the infused samples, in agreement with what has previously been shown both on infused polymers[8,9] and other types of IL interfaces.[4,10]
Figure 4. Comparison of S. aureus and P. aeruginosa attachment on PDMS and infused PDMS after 48 h under dynamic conditions. (A) Biofilm stain absorbance (B) Counts normalized to the average non-infused PDMS values.
The normalized removal curves shown in Figure 4 exhibit drastically different results for both S. aureus and P. aeruginosa grown under dynamic conditions on infused samples compared to what was observed for E. coli subjected to the same treatment. For S. aureus, the first removal cycle was found to remove nearly all the adherent bacteria (������ = 0.60), in a manner more similar to E. coli grown under static conditions. However, subsequent removal cycles revealed that the few bacteria that remained behind appeared to be even less likely to be removed than the E. coli (������ = 0.44), with bacteria still being removed from the surface after the 4th cycle. Tests on P. aeruginosa, in contrast, revealed an intermediate decrease in
13

CFUs removed during the first cycle (������ = -0.06), but much less persistent attachment in the following cycles (������ = 2.00), and a drop to 0 CFU by the third cycle.
The removal curves for S. aureus and P. aeruginosa shown in Figure 5B reveal an interesting difference in relative adhesion to infused surface under dynamic conditions, especially in contrast to E. coli. The presence or absence of flagella may partially explain the observed results, as they were shown in previous experiments on textured superhydrophobic surfaces to play a definitive role in adhesion.[30] E. coli have multiple long flagella per cell; in contrast, P. aeruginosa have a single flagellum per cell, while S. aureus have none. The decreasing normalized CFU values for the first removal cycle reflect this (0.85, 0.39, and 0.15 for E. coli, P. aeruginosa, and S. aureus, respectively). These results may suggest that, to some extent, the relative strength of initial adhesion of a particular bacterial species to infused surfaces under dynamic conditions may be predicted by the number of flagella.
It is also clear from the subsequent removal cycles for all three species that other factors in addition to flagella are likely contributing to the interaction with the infused polymer surface and/or liquid overlayer. These may include cell shape and surface charge,[31] cell membrane composition,[32] and/or cell motility.[33] Now that the methods presented here have been shown effective at revealing the differences between different species, further tests should include a systematic investigation of these factors as well.
2.4. Effect of Flagella To further investigate the question of the relative importance of flagella in adhesion to
the IL layers on our oil-infused samples, we tested E. coli with the flagellin subunit gene deleted (ΔfliC mutant)[15], as well as the wild type (WT) K12 strain from which it had been derived as a control. Previous work has revealed a critical role for flagella in the adhesion of E. coli to surfaces.[14,15]
14

The normalized and fitted CFU count results for E. coli with and without flagella are shown in Figure 4A. The results for the WT control show a pattern similar to that of the clinical isolate previously tested, with no significant decrease in removed CFUs upon the first removal cycle (������ = -1.37, P = 0.006) and a similar degree of removal for the subsequent cycles (������ = 0.99), becoming significantly different from the control upon the 3rd removal cycle (P = 0.022). The degree of similarity observed in the removal curves between this WT K12 strain and the clinical isolate tested in Figure 3 further validate both the method and the fit as an approach to accurately quantify the relative degree of bacterial adhesion on infused PDMS surfaces.
The removal curve for the bacteria without flagella (E. coli ΔfliC) grown on IL layers under dynamic conditions shows a clear difference from the WT control (Figure 5A). Although the amount of bacteria removed from the surface in the first removal cycle was not as great as that of the samples grown under static conditions (������ = -0.12 compared to ������ = 0.96, respectively), the degree of removal in subsequent cycles was much higher (������ = 3.78 for E. coli ΔfliC compared to ������ = 0.74 for E. coli grown in static conditions). These results suggest that for the E. coli straing tested here, the flagella play an important role in adhesion to the oil-infused surfaces under dynamic conditions.
15

Figure 5. Adhesion of E. coli Wild Type (WT, with flagella) versus the ΔfliC mutant (without flagella) on non-infused and infused PDMS. (A) CFU counts normalized to control PDMS values and (B) a mixing assay showing the stability of an emulsion of the infusing liquid (silicone oil) with a bacterial solution.
To further test the effects of both the presence and absence of flagella on interaction with the silicone oil present in our infused materials, we performed a mixing assay[15] in which a volume of silicone oil was vortexed with an equal volume of medium containing bacteria either with or without flagella (Figure 5B). This experiment was intended to confirm that the absence of flagella was indeed changing the way that the bacteria interacted with the oil itself. It was found that while both types of bacteria would initially produce an emulsion, the sample containing bacteria without flagella would separate after 24 h while the sample containing bacteria with flagella would remain as a stable emulsion. These results support what was observed in the removal curves – that E. coli without flagella are less likely to have stable interactions with the silicone oil.
One explanation for the observation that flagella contribute to adhesion to infused surfaces, but do not fully account for it may be the composition of the E. coli cellular
16

membrane. The membrane is known to contain proteins with hydrophobic domains that could provide stabilization when in contact with the oil layer.[34] However, stabilization with the membrane alone does not appear to be as effective as stabilization with flagella, as flagella not only contain hydrophobic domains and can act as a surfactant, but also have the surface area and length to potentially penetrate down into the oil layer or spread out along the surface to improve contact. The fact that greater adhesion occurs most strikingly for samples grown under dynamic conditions may also tie in to the observation that movement of the medium appears to disrupt the oil overlayer. If the layer is deformable, there would be more opportunity for the flagella to interact with it and become stabilized. Further work should focus on the localization of the E. coli flagella with respect to the oil overlayer.
3. Conclusion The results presented here show that the interaction of bacteria with IL layers is a
complex phenomenon involving the integrity of the oil layer itself and the morphology of the interacting cell. Importantly, in all cases, we show that bacterial attachment and biofilm formation on oil-infused samples were significantly reduced compared to non-infused controls, and that the removal of the attached cells was easier and required a smaller number of strip-off cycles. Within the set of oil-infused samples, we find that growing E. coli under dynamic orbital flow conditions initially results in the adhesion of a greater number of cells compared to growing E. coli under static conditions. Interestingly, this increase of attached cells does not correlate with an equal increase in biofilm formation. We further show that number of adherent cells of two more clinically relevant species, S. aureus and P. aeruginosa, is much lower than that of E. coli grown under similar dynamic conditions, suggesting a species-dependence to the ability of IL layers to resist bacterial attachment. Finally, we demonstrate that these differences may be due in part to the presence of flagella, as flagella-
17

deficient E. coli mutants adhered significantly less well under dynamic conditions than their WT counterparts.
Overall, our work shows that IL layers of silicone oil on infused PDMS are capable of reducing bacterial adhesion under dynamic conditions, even though the introduction of dynamic conditions affects the state of the IL layer. These results also support previous evidence that adhesion to IL layers can be species-, and even strain-dependent,[10] illuminating the need for further research into the interaction of IL layers and cell surface structures, such as adhesins, flagella, and fimbriae. An increased understanding of these interactions will allow for the modification of existing IL systems to better resist the adhesion of a broader range of microorganisms.
4. Experimental Section Bacterial Stock Preparation: The wild-type (WT) bacterial strains cultured were E. coli strain J96, E. coli strain K12 W3110, P. aeruginosa, and S. aureus. The genetically altered cultures used were E. coli strain MC4100 mKate2::KanR and E. coli strain K12 W3110 ΔfliC. All stocks were generated from a stock agar plate, from which three to five colonies were transferred into growth media. All cultures were incubated overnight at 37 oC in the growth media, LB Miller Broth (Becton, Dickinson and Company, Franklin Lakes, NJ), and kept shaking continuously in an orbital shaker. The mutant culture E. coli mKate2 was selected with 50 µg mL-1 of kanamycin sulfate (Life Technologies Corporation, Chicago, IL).
PDMS and Infused PDMS Preparation: Fluorescent PDMS stock was prepared by dissolving 850 µg of BODIPY® FL alkyne laser dye (Exciton, Dayton, OH) into 10 ml of dichloromethane (Sigma Aldrich, St. Louis, MO), then mixing with 100 g of Sylgard® 184 silicone elastomer base (Dow Corning Corporation, Midland, MI). After mixing and degassing under vacuum overnight, the stock of fluorescent PDMS base was combined with
18

its respective Sylgard® curing agent in a 10:1 ratio and homogenized in a Thinky planetary centrifugal mixer (Thinky Corporation, Tokyo, Japan) at 2000 rpm for 1 minute, then again at 2200 rpm for 1 minute. Aliquots of 0.9 mL were then transferred into the square depressions (2.0 cm x 2.0 cm x 0.5 cm) of a mold master. The master was kept under vacuum for 2 h to extract air trapped in the PDMS, then at 70 oC overnight to facilitate PDMS curing. The cured PDMS squares were removed from the mold, cut in half, and sterilized in a glass dish via autoclave without a drying cycle. Infused PDMS samples were created by aseptically immersing the autoclaved PDMS coupons in 10 cSt silicone oil (Gelest, Morrisville, PA), previously saturated overnight with BODIPY® PM597-8C9 laser dye (Lumiprobe, Hallandale Beach, FL) and filtered through a 0.22 µm nylon syringe filter (Sigma Aldrich, St. Louis, MO).
Biofilm Growth: Difco tryptic soy broth (Becton, Dickinson and Company, Franklin Lakes, NJ), made with distilled, deionized water at 30 mg mL-1 and containing an additional 1.5% (w/v) NaCl (Sigma Aldrich, St. Louis, MO), was used to induce E. coli and S. aureus biofilms. 25 mg mL-1 LB Miller Broth + 0.2 % sodium citrate (w/v) (Sigma Aldrich, St. Louis, MO) made with distilled, deionized water was used to induce P. aeruginosa biofilms. Non-infused and infused PDMS samples were placed into 6-well Falcon plates (Fisher Scientific, Pittsburgh, PA), and each well was filled with 7 mL of biofilm inducing media and 70 µL of bacterial culture (1:100 inoculum to media). Kanamycin was added at 50 µg mL-1 for selection of E. coli mKate2 prior to inoculating media. Triplicates were kept for both infused samples and their non-infused controls. Immediately following inoculation, all samples were incubated at 37 oC for 48 h to allow for proper biofilm formation. To simulate dynamic conditions during this period, one set of samples was kept shaking continuously at 100 rpm via placement on an orbital shaker inside the incubator. This speed value was chosen as it was the highest that could be achieved without spilling the culture medium out of the containers.
19

Crystal Violet Staining and Analysis: Biofilm coverage was analyzed by staining samples via a 10-minute incubation in a 0.1% (w/v) solution of crystal violet (CV) in deionized, distilled water. This was followed by gentle rinsing in deionized, distilled water to remove excess stain, air-drying, and photography with a Canon EOS Rebel T4i camera (Melville, NY). After image capture, the samples were placed into 15 mL Falcon tubes (Fisher Scientific, Pittsburgh, PA) and CV was dissolved away in 5 to 7 mL of 70 % ethanol (v/v) in deionized, distilled water. 200 µL of the resultant solution was transferred in triplicate into a 96 well Falcon plate (Fisher Scientific, Pittsburgh, PA). Endpoint absorbance at 590 nm was measured using a Spectramax i3 - UV/VIS spectrometer (Molecular Devices, Sunnyvale, CA). Absorbance was normalized to the product of the volume of ethanol and surface area exposed to biofilm formation.
Image analysis was performed as following using MATLAB (Mathworks, MA, USA). Photographs taken of the CV-stained samples were cropped to the region of interest (i.e. to substrate edges) and converted from RGB color model to HSV (hue, saturation, value). Cropped images (n = 3) were binarized based on an interval defined in hue/saturation/value and the biofilm coverage was defined as percentage of the binarized biofilm over the whole image area.
Adhesion Assay and Relative Adhesion Analysis: PDMS/infused PDMS samples were removed from culture aseptically and immediately dipped into a 15 mL Falcon tube (Fisher Scientific, Pittsburgh, PA) containing 10 mL of sterile 1X phosphate buffer solution (PBS) to remove planktonic bacteria. There was no clear difference in CFU counts of these planktonic bacteria between static and dynamic growth conditions, although the results were highly variable (1.7x107 ± 3.7 x107 cells/mL-cm2 for samples grown under static conditions and 3.1x106 ± 2.2 x106 cells/mL-cm2 for samples under dynamic conditions). Each sample was
20

then transferred to another 15 mL Falcon tube containing 10 mL of sterile 1X PBS and

subsequently vortexed for 90 s at the highest setting (level 10) on a VWR Analog Vortexer

(VWR, Radnor, PA) and sonicated for 180 s in a Branson B8510 Ultrasonic cleaner (Branson

Ultrasonics Corp., Danbury, CT). One round of vortexing and sonication for these durations

was considered one removal cycle, thought to reliably remove bacteria adherent to the

surface.[27] This was repeated with all samples for a total of 4 cycles. Removal supernatant

was serially diluted from 100 to 10-6 in 1.7 mL mini-centrifuge tubes (VWR, Radnor, PA). 10

µL from each dilution was plated onto a LB 1.5 % agar plate. Plates were incubated at 37 oC

for 12 h. Colonies were counted and colony forming unit (CFU) count in CFU mL-1 cm-2 was

calculated. When plotting normalized CFU data, each removal fraction mean count was

normalized

to

the

non-infused

sample

mean

count

and

plotted

as

5678 5679

±

������������.

������.

where CFUi represents the CFU count of the infused sample, CFUc the CFU count of the

corresponding control sample, and St.E the standard error.

Confocal Imaging: Aseptically removed PDMS samples of dimensions 2 cm x 1 cm x 0.5 cm were dipped once into 1X PBS to remove planktonic organisms. Samples were moved into 60 mm x 15 mm circular petri dishes (VWR, Radnor, PA), then 1X PBS was pipetted into each dish until the sample was submerged. Confocal images were captured using a 40x/1.0 water dipping objective lens on an Upright Zeiss LSM 710 (Zeiss, Thornwood, NY).

Statistics: One-way analysis of variance (ANOVA) tests were used to test the hypotheses for
image analysis and biofilm staining. Comparison of CFU counts were performed using square
root-transformed data in a univariate ANOVA to compare between non-infused and infused
counts at a given removal stage, with P < 0.05 considered significant.
Supporting Information Supporting Information is available from the Wiley Online Library or from the author.
21

Acknowledgements Y. Kovalenko and I. Sotiri contributed equally to this work. The authors thank Ronn Friedlander for providing the E. coli ΔfliC mutant, Phil Kim for helpful advice, and Abigail Weigang for editing assistance. This material is based upon work supported by the Defense Advanced Research Projects Agency Grant N66001-11-1-4180 and Contract HR0011-13-C0025.

References

Received: ((will be filled in by the editorial staff)) Revised: ((will be filled in by the editorial staff))
Published online: ((will be filled in by the editorial staff))

[1] J. Guggenbichler, O. Assadian, M. Boeswald, A. Kramer, GMS Krankenhaushygiene

Interdiszip. 2011, 6, 1.

[2] G. Shepherd, P. Mohorn, K. Yacoub, D. W. May, Ann. Pharmacother. 2012, 46, 169.

[3] T.-S. Wong, S. H. Kang, S. K. Y. Tang, E. J. Smythe, B. D. Hatton, A. Grinthal, J.

Aizenberg, Nature 2011, 477, 443.

[4] A. K. Epstein, T.-S. Wong, R. a Belisle, E. M. Boggs, J. Aizenberg, Proc. Natl. Acad.

Sci. U. S. A. 2012, 109, 13182.

[5] I. Sotiri, J. C. Overton, A. Waterhouse, C. Howell, Exp. Biol. Med. 2016, in press, 1.

[6] D. C. Leslie, A. Waterhouse, J. B. Berthet, T. M. Valentin, A. L. Watters, A. Jain, P.

Kim, B. D. Hatton, A. Nedder, K. Donovan, E. H. Super, C. Howell, C. P. Johnson, T.

L. Vu, D. E. Bolgen, S. Rifai, A. R. Hansen, M. Aizenberg, M. Super, J. Aizenberg, D.

E. Ingber, Nat. Biotechnol. 2014, 32, 1134.

[7] B. Solomon, K. Khalil, K. Varanasi, Bull. Am. Phys. Soc. 2013, 30, 10970.

[8] C. Howell, T. L. Vu, J. J. Lin, S. Kolle, N. Juthani, E. Watson, J. C. Weaver, J.

Alvarenga, J. Aizenberg, ACS Appl. Mater. Interfaces 2014, 6, 13299.

[9] N. MacCallum, C. Howell, P. Kim, D. Sun, R. Friedlander, J. Ranisau, O. Ahanotu, J. J. 22

Lin, A. Vena, B. Hatton, T. Wong, J. Aizenberg, ACS Biomater. Sci. Eng. 2015, 1, 43. [10] J. Li, T. Kleintschek, A. Rieder, Y. Cheng, T. Baumbach, U. Obst, T. Schwartz, P. A.
Levkin, ACS Appl. Mater. Interfaces 2013, 5, 6704. [11] G. O. Toole, H. B. Kaplan, R. Kolter, Annu. Rev. Microbiol. 2000, 54, 49. [12] A. Persat, C. D. Nadell, M. K. Kim, F. Ingremeau, A. Siryaporn, K. Drescher, N. S.
Wingreen, B. L. Bassler, Z. Gitai, H. A. Stone, Cell 2015, 161, 988. [13] T. Vanzieleghem, P. Herman-Bausier, Y. F. Dufrene, J. Mahillon, Langmuir 2015, 31,
4713. [14] R. Van Houdt, C. W. Michiels, 2005, 156, 626. [15] R. S. Friedlander, N. Vogel, J. Aizenberg, Langmuir 2015, 31, 6137. [16] G. a. O’Toole, R. Kolter, Mol. Microbiol. 1998, 30, 295. [17] J. M. Willey, L. M. Sherwood, C. J. Woolverton, in Prescott’s Microbiol., McGraw-
Hill, New York, 2011, pp. 183–185. [18] W. E. Thomas, L. M. Nilsson, M. Forero, E. V. Sokurenko, V. Vogel, Mol. Microbiol.
2004, 53, 1545. [19] S. Lecuyer, R. Rusconi, Y. Shen, A. Forsyth, H. Vlamakis, R. Kolter, H. A. Stone,
Biophys. J. 2011, 100, 341. [20] C. J. Kavanagh, G. W. Swain, B. S. Kovach, J. Stein, C. Darkangelo-Wood, K. Truby,
E. Holm, J. Montemarano, A. Meyer, D. Wiebe, Biofouling 2003, 19, 381. [21] K. Truby, C. Wood, J. Stein, J. Cella, J. Carpenter, C. Kavanagh, G. Swain, D. Wiebe,
D. Lapota, a Meyer, E. Holm, D. Wendt, C. Smith, J. Montemarano, Biofouling 2000, 15, 141. [22] C. Howell, T. L. Vu, C. P. Johnson, X. Hou, O. Ahanotu, J. Alvarenga, D. C. Leslie, O. Uzun, A. Waterhouse, P. Kim, M. Super, M. Aizenberg, D. E. Ingber, J. Aizenberg, Chem. Mater. 2015, 27, 1792. [23] U. Manna, N. Raman, M. A. Welsh, Y. M. Zayas-Gonzalez, H. E. Blackwell, S. P.
23

Palecek, D. M. Lynn, Adv. Funct. Mater. 2016, 3599. [24] K. Otto, T. J. Silhavy, Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 2287. [25] R. Belas, Mol. Microbiol. 2013, 90, 1. [26] R. Belas, Trends Microbiol. 2014, 22, 517. [27] G. Bjerkan, E. Witsø, K. Bergh, Acta Orthop. 2009, 80, 245. [28] J. F. Douglas, H. E. Johnson, S. Granick, Science (80-. ). 1993, 262, 2010. [29] F. Song, D. Ren, Langmuir 2014, 30, 10354. [30] R. S. Friedlander, H. Vlamakis, P. Kim, M. Khan, R. Kolter, J. Aizenberg, Proc. Natl.
Acad. Sci. U. S. A. 2013, 110, 5624. [31] J. Feldner, W. Bredt, I. Kahane, J. Bacteriol. 1983, 153, 1. [32] A.-L. Henche, A. Koerdt, A. Ghosh, S.-V. Albers, Environ. Microbiol. 2012, 14, 779. [33] H. Morisaki, S. Nagai, H. Ohshima, E. Ikemoto, K. Kogure, Microbiology 1999, 145,
2797. [34] L. S. Dorobantu, A. K. C. Yeung, J. M. Foght, M. R. Gray, Appl. Environ. Microbiol.
2004, 70, 6333.
The number of bacterial cells adherent on an immobilized liquid layer of silicone oil is significantly affected by the presence or absence of orbital flow, the species of bacteria, and the bacterial cell morphology. In all tested experimental conditions, however, immobilized liquid layers show superior resistance to irreversible bacterial adhesion compared to equivalent solid polydimethylsiloxane (PDMS) surfaces.
Keywords: Infused polymers, IL layers, biofilm, bacteria, infection, SLIPS Y. Kovalenko, I. Sotiri, J. V. I. Timonen, J. C. Overton, G. Holmes, J. Aizenberg, C.Howell* Bacterial Interactions with Immobilized Liquid Layers ToC figure
24

25

Copyright WILEY-VCH Verlag GmbH & Co. KGaA, 69469 Weinheim, Germany, 2013. 26