On-Demand Removal of Bacterial Biofilms via Shape Memory Activation.

Bacterial biofilms are a major cause of chronic infections and biofouling; however, effective removal of established biofilms remains challenging. Here we report a new strategy for biofilm control using biocompatible shape memory polymers with defined surface topography. These surfaces can both prevent bacterial adhesion and remove established biofilms upon rapid shape change with moderate increase of temperature, thereby offering more prolonged antifouling properties. We demonstrate that this strategy can achieve a total reduction of Pseudomonas aeruginosa biofilms by 99.9% compared to the static flat control. It was also found effective against biofilms of Staphylococcus aureus and an uropathogenic strain of Escherichia coli.

Bacteria can colonize both biotic and abiotic surfaces and form biofilms that are multicellular structures with extracellular polymeric substrates secreted by the attached cells.[1] Biofilm cells are difficult to eradicate compared to their planktonic counterparts due to enhanced resistance to antimicrobials and mechanical forces.[2] The grand challenge of biofilms has motivated the search for new strategies for biofilm prevention and removal.[3]

In the past decade, surface topography has been well studied for its potential in controlling bacterial adhesion and biofilm formation.[4−9] For example, topographic lines, static nanohair like structures and patterns arranged in diamond arrays have been embossed onto poly(dimethylsiloxane) (PDMS) surfaces to mimic cilia and the topographies on shark skin, respectively,[7−9] for biofilm control. In addition to static features of surface chemistry and topography, bacterial adhesion can also be prevented by stimuli-responsive antifouling surfaces, such as those triggered by pH, temperature, salt concentration, electrical potential, light, magnetic field, and the surrounding media.[10−14] Movements of or near the substrate surfaces such as these provide an additional level of control that can disperse attached cells.

Inspired by the natural systems and previous research,[15] we aimed to engineer novel antifouling surfaces that can be programmed to remove well-established biofilms. We hypothesized that by changing the shape and dimension of topographic patterns, the biofilm structure can be disrupted, leading to biofilm dispersion. To test this hypothesis, we evaluated the effects of dynamic change in surface topography on biofilms formed on shape memory polymer (SMP) surfaces with recessive hexagonal patterns. SMPs are a class of materials that can memorize a permanent shape through physical or chemical cross-linking, be manipulated and fixed in a temporary shape via an immobilizing transition, such as vitrification or crystallization, and subsequently recover to the permanent shape as the result of a triggering event, such as thermal, electrical, light, or solvent activation.[16,17] This phenomenon is known as one way shape memory, because activity occurs in one direction. In addition to one way SMPs, there are also two way, triple shape, multishape, and multifunctionality SMPs,[18] which make SMP with systematically designed surface topography a promising candidate for the development of novel and programmable antifouling strategies. We chose hexagonal patterns because the static protruding or recessive hexagonal patterns have been found to reduce biofilm formation significantly.[4,5] Thus, they are good candidates for biofouling control using dynamic topography in this study. Another advantage to use recessive hexagonal patterns is that they can maintain structural integrity under a uniaxial strain of >50%, an important step in creating the temporary shape. Pseudomonas aeruginosa, Staphylococcus aureus, and Escherichia coli were used as model species in this study due to their significant roles in biofilm infections.[19]

In this proof-of-concept study, we chose an SMP based on t-butyl acrylate (tBA), poly(ethylene glycol) dimethacrylate (PEGDMA), and photoinitiator 2,2-dimethoxy-2-phenylacetophenone (DMPA), which has one way shape memory around its glass transition temperature (Tg; Figure S1).[20] The biocompatibility of this SMP has been validated by its cardiovascular applications.[21] Using dynamic mechanical analysis (DMA), we confirmed that this SMP has a stepwise decrease in tensile storage modulus at 47.5 °C, with a glassy modulus of about 700 MPa below this temperature and a rubbery plateau of about 1.3 MPa above this temperature (Figure S2), which suggests that this SMP can be deformed at temperatures above 47.5 °C and fixed at temperature below 30 °C. The effective Tg in 0.85% (wt/vol) NaCl solution was found to be 38.5 °C, which is lower than the dry Tg (44.3 °C) due to plasticization by water (Figure S3), allowing rapid shape change near body temperature (37 °C).[20] These shape memory properties of our material are consistent with literature values.[20] Stretched SMP surfaces used in this study were found to stably maintain their fixed temporary shape during 48 h incubation at room temperature in sterile LB medium. After 10 min of incubation at 40 °C, the programed SMP shank with a 98.9% recovery to the permanent shape (Figure S1).

Static flat control (prepared without shape memory fixing so as not to change shape when heated to 40 °C) and both flat and topographically patterned programmed surfaces (fixed with a temporary but stable uniaxial strain of >50% so as to contract by ∼50% when heated to 40 °C) were prepared (Figure S4). All surfaces were challenged with biofilm formation of P. aeruginosa PAO1, S. aureus ALC2085, and uropathogenic E. coli ATCC53505 for 48 h at room temperature.

We first studied the effects of static topography on adhesion and biofilm formation by comparing the biomass of 48 h biofilms formed on these three different surfaces. For calculating biomass, 3D information was obtained from a series of z stack biofilm images (1 μm interval), which were then analyzed using the software COMSTAT.[22] By analyzing the biomass of 48 h biofilms on these three different surfaces, hexagonal recessive patterns were found to significantly reduce microbial biofilm formation. For example, the biomass of P. aeruginosa PAO1 biofilms on topographically patterned programmed substrates (before triggered shape recovery) was 50.9 ± 7.2% and 51.9 ± 7.3% of that on flat programmed substrates and static flat control, respectively (p < 0.001 for both, one way ANOVA adjusted by Tukey test; Figure a). No significant difference was found between static flat controls and flat programmed substrates (both around 9 μm3/μm2; p = 0.93 one way ANOVA).

Figure 1

Biofilm formation of P. aeruginosa PAO1 on static flat control and programmed substrates (both flat substrates and substrates patterned with 10 μm deep recessive hexagonal patterns) fixed with a temporary but stable uniaxial strain of >50% so as to contract by ∼50% when heated to 40 °C. The figures show the biomass (a) and representative fluorescence images (b) of P. aeruginosa PAO1 biofilms on different surfaces before and after trigger (10 min incubation at 40 °C) (bar = 50 μm). Mean ± standard deviation shown.

After 48 h of incubation, we tested the effects of topographic changes on established biofilms. Upon heating for 10 min at 40 °C, shape recovery induced significant detachment of established biofilms. For example, the biomass on topographically patterned programmed substrates was 4.7 ± 0.7 μm3/μm2 and 0.01 ± 0.01 μm3/μm2 before and after shape recovery induced changes in surface topography, respectively. This represents a 469-fold reduction of biomass due to the change in substrate topography, and 909-fold reduction comparing to the 48 h biofilm biomass (9.1 ± 0.8 μm3/μm2) on static flat controls without topographic patterns and shape change. Collectively, these data demonstrate up to 99.9% biofilm reduction through combined effects of biofilm inhibition by surface topography and biofilm removal by shape change. Similar effects of biofilm removal were also observed for flat programmed substrates, e.g., the biomass on flat programmed substrates was 9.3 ± 2.9 μm3/μm2 and 0.04 ± 0.03 μm3/μm2 before and after shape recovery, respectively (231-fold reduction, p < 0.001, one way ANOVA adjusted by Tukey test) (Figure a). These results were corroborated by fluorescence images (Figure b) and colony forming unit (CFU) assay (Figure S5).

In contrast to the reduction in biomass observed on programmed substrates, the biomass on static flat controls before and after incubation at 40 °C for 10 min was 9.1 ± 0.8 μm3/μm2 and 8.5 ± 1.9 μm3/μm2, respectively (p = 0.63, one way ANOVA; Figure a). Thus, the aforementioned biofilm removal was indeed caused by shape change rather than temperature change.

Biofilm dispersion was further verified by taking real-time movies. Before triggered shape change by heating for 10 min at 40 °C, P. aeruginosa PAO1 biofilms were clearly seen with large cell clusters (Movies S1 and S2). When the shape recovery started, rapid detachment of both cell clusters and individual cells were observed on both flat and topographically patterned programmed substrates (Movies S1 and S2). Most changes in shape occurred in the first 6 min after shape recovery started (Figure a,b). Surface coverage by biofilms was 33.0% before shape recovery (t = 0 s) and dropped to 19.9% after just 4.3 s of shape recovery (Figure b). At 6 min, surface coverage further decreased to 11.1% (Figure b). It is worth noticing that this experiment was conducted without flow, and a gentle wash after shape change was sufficient to remove nearly all detached cells (Figures and 2c). Such detachment was not observed for the static flat control (no shape recovery, Movie S3), which was also incubated at 40 °C for 10 min (Figure d). After 10 min shape recovery, the same cell clusters remained on these static control surfaces (Figure d).

Figure 2

Biofilm removal during shape change. (a) A 3D image of P. aeruginosa PAO1 biofilm detachment. This 3D image was taken when the rapid biofilm detachment occurred in the first 4.3 s after topographic transition started. Due to the fast cell movement, trajectories of detached cells and cell clusters were recorded as the z stage moved upward (representative cells highlighted using white arrows). (b) Length and width of recessive hexagonal patterns measured during topographic change and the surface coverage of P. aeruginosa PAO1 biofilms at 0, 4.3, 360, and 600 s after the beginning of shape recovery and the final surface after washing. (c and d) Fluorescence images of P. aeruginosa PAO1 biofilms on topographically patterned programmed substrates (c) and static flat control (d) during triggered shape change (10 min incubation at 40 °C) (bar = 50 μm). Images show that cell clusters were removed from the patterned SMP with shape change but remained on the flat control surfaces.

The biocompatibility of this SMP chemistry has been demonstrated by its cardiovascular application;[21] however, the toxicity of this SMP to bacterial cells has not been evaluated. To test if this SMP is toxic to bacterial cells, we grew planktonic P. aeruginosa PAO1 cells in the presence of 0, 1, 5, and 10% (wt/vol) of this SMP. The planktonic growth of P. aeruginosa PAO1 in the presence of SMP was not significantly different than that of cells in LB medium (p > 0.05, one way ANOVA), indicating that the SMP used this study is not toxic to P. aeruginosa PAO1 (Figure a).

Figure 3

SMP is not toxic to P. aeruginosa PAO1 cells. (a) Growth curves of P. aeruginosa PAO1 in the presence of different amounts of SMP (0, 1, 5, or 10% wt/vol). (b) Effect of 10 min incubation at different temperatures (37, 38, 39, 40, 41, or 42 °C) on the viability of P. aeruginosa PAO1 cells. Mean ± standard deviation shown.

We further evaluated the effects of temperature change (10 min at 40 °C) on the viability of P. aeruginosa PAO1. By quantifying the number of viable cells after 10 min incubation at temperatures ranging from 37 to 42 °C, we found that the viability of P. aeruginosa PAO1 cells was not affected by any of the tested conditions (Figure b; p > 0.05, one way ANOVA). These and the above results confirmed that the reduction in biofilm biomass was due to biofilm dispersion by shape recovery rather than the toxicity of SMP or thermo-induced killing.

To understand if the effects of dynamic topography are species specific, we repeated the biofilm experiment using S. aureus and an uropathogenic E. coli strain. Similar to the results of P. aeruginosa PAO1, topographically patterned programmed substrates exhibited 39.8 ± 4.0% inhibition of 48 h E. coli biofilm formation compared to the static flat control (p < 0.001, two way ANOVA adjusted by Tukey test; Figure S6) before triggered shape change. No significant difference (both around 3.3 μm3/μm2 (Figure S6a); p = 0.73, one way ANOVA) was found between static flat control and flat programmed substrates. The biomass of 48 h S. aureus biofilms on topographically patterned programmed substrates was similar to that on static flat control and flat programmed substrates (both around 5.5 μm3/μm2; p = 0.22, one way ANOVA; Figure S7a). Nevertheless, change in surface topography triggered by shape recovery still caused dramatic detachment of both E. coli and S. aureus biofilms (Figures S6a and S7a). For example, the biomass of S. aureus biofilms on topographically patterned programmed substrates was 5.5 ± 0.2 μm3/μm2 and 0.04 ± 0.02 μm3/μm2 before and after shape recovery, respectively (Figure S7a). Similar effects of biofilm removal were also observed for flat programmed substrates (Table S1). In contrast, incubation at 40 °C for 10 min alone did not show significant effect on the biofilms formed on flat control substrates, showing that biofilm removal from stretched samples was indeed caused by shape recovery. Some representative fluorescence images are shown in Figures S6b and S7b.

To understand the long-term effects of biofilm removal and how fast the remaining cells can reform biofilms, we followed the regrowth of P. aeruginosa PAO1 and E. coli ATCC53505 biofilms at 12, 24, and 48 h after shape recovery trigged biofilm removal, and compared the results with the biomass before shape recovery and the static flat control that underwent the same temperature change but not shape recovery. The data summarized in the Figures S8 and S9 show that, for both species, the biomass on the surfaces that had gone through shape recovery was significantly lower than that before shape recovery and on the static flat control. For example, the biomass of P. aeruginosa PAO1 biofilms on flat program surfaces was 1.6 ± 0.1 μm3/μm2 at 48 h after shape recovery. This is 83.7% (p = 0.0164, one way ANOVA adjusted by Tukey test, Figure S8) lower than that before shape change (9.3 ± 2.9 μm3/μm2) and 89.0% (p = 0.0022, one way ANOVA adjusted by Tukey test, Figure S8) lower than that on the static flat control (increased from 9.1 ± 0.8 μm3/μm2 to 14.7 ± 0.9 μm3/μm2 during the same period of incubation time). Even stronger effects were found for patterned programed surfaces (additional 58.1% reduction than the above flat programed surfaces; p = 0.0002, one way ANOVA adjusted by Tukey test, Figure S8) and consistent results were obtained for E. coli ATCC53505 biofilms (Figure S9). Collectively, these results indicate that the biofilm regrowth after shape changes was relatively slow (83.7% and 85.8% less biofilm after 48 h of regrowth compared the biomass after 48 h of the initial biofilm formation on new flat and patterned programmed surfaces, respectively), presumably because of the mass reduction of biofilm biomass by shape recovery.

Despite the extensive research on fouling control during the past decades,[23,24] biocompatible materials that offer long-term biofilm control in complex environment are still yet to be developed. Moreover, removing mature biofilms that have large cell clusters and thick extracellular matrices remains as an unmet challenge. In this study, we introduced recessive hexagonal patterns on SMP substrates to inhibit biofilm formation and obtained dynamic change in surface topography upon triggered shape memory recovery. The shape-change-induced biofilm dispersion was fast (∼6 min) and can remove large clusters from mature biofilms. This material is also biocompatible,[21] and the shape change can be triggered by gentle heating, without using an electric or magnetic field as required by some other systems.[12,25]

The topography was created using soft lithography;[26] thus, it is well-defined and can be applied to a large surface area. Despite these advantages, we are aware that this SMP only has one way shape change. To be broadly adapted for diverse applications, the capability to go through cyclic changes in shape is desirable. Some shape memory polymer chemistries have been demonstrated to have two way, triple shape, or other forms of multi shape.[18,27−29] In the future, we plan to test such polymers to obtain more sustainable antifouling properties. It will also be helpful for biomedical applications to have the temporary shape maintained at body temperature rather than room temperature. This is part of our ongoing study. With regards to the mechanism of biofilm dispersion, data presented herein (Movies S1 and S2) revealed that biofilm dispersion was rapid and cell clusters were disrupted. The exact mechanism of shape memory recovery triggered biofilm removal is unknown. We speculate that the observed effects might be caused by disruption of biofilm matrix and cell–surface interactions. This is part of our ongoing work.

In summary, we developed new antifouling surfaces based on shape memory triggered changes in surface topography. This strategy was found effective for the removal of established biofilms of multiple species. Future studies are needed to understand the underlying mechanism and develop biocompatible polymers for in vivo use. Long-term biofilm control may be possible by employing surface topographies on such polymers to achieve biofilm inhibition and self-cleaning.