On-demand biofilm removal by shape-memory triggered local changes in surface topography

TL;DR

Researchers developed a method to remove bacterial biofilms from medical devices using shape-memory materials that change surface texture without altering the device's shape.

Contribution

A novel strategy for biofilm removal using localized surface topography changes in shape-memory polymers, without altering the bulk material shape.

Findings

Microscale surface topography changes triggered by shape-memory polymers removed up to 71.5% of Pseudomonas aeruginosa biofilms.

The method increased antibiotic susceptibility of remaining biofilm cells.

Surface topography changes occurred without altering the bulk material's shape.

Abstract

Bacterial pathogens can form biofilms on implanted biomedical devices, causing persistent infections that are highly tolerant to antibiotics. Previously, we reported a strategy of biofilm control based on dynamic topography, which effectively removes biofilms via horizontal contraction of the substrate surface of a shape-memory polymer (SMP) upon triggered shape recovery. This method is effective and species nonspecific; however, alterations in the bulk material profile limit its applications. In this study, we tested the hypothesis that biofilm can be removed by changes in local topography without altering the shape of the bulk material. Acrylate-based SMPs were prepared to obtain transition temperature of 40℃ to trigger shape recovery in aqueous environment within 10 min. Micron-scale square patterns that are about 6-µm tall with varying width and spacing were prepared by hot compression against PDMS with complementary patterns, while maintaining the bulk shape of the material unchanged. The results demonstrated effective on-demand biofilm removal (e.g., 48 h biofilms of Pseudomonas aeruginosa and 24 h biofilms of Escherichia coli were removed by 71.5% and 70.6%, respectively). In addition, shape recovery triggered topographic changes increased antibiotic susceptibility of attached bacterial cells. Overall, the results from this study demonstrated the feasibility to remove biofilms without changing the shape of the bulk material. These findings are helpful for engineering better antifouling materials.

Bacterial biofilms are the root cause of persistent infections associated with implanted biomaterials. Conventional treatments with antibiotics are often ineffective and promote the development of bacterial drug resistance. Thus, we are motivated to engineer new biomaterials that are self-defensive against bacterial colonization. Previously, we reported that shape-memory polymers (SMPs) can be programed to change the bulk shape (via horizontal stretch) on-demand and effectively remove bacterial biofilms. In this study, we further developed this strategy to control shape change of surface topography alone. The SMP surfaces programmed with microscale square-shaped features were fabricated, which were able to revert to flat surfaces upon triggering with moderate temperature change and disrupt bacterial biofilms (~70%). The shape recovery was limited to surface topography with the bulk shape unchanged. In addition to biofilm removal, shape recovery also enhanced the antibiotic susceptibility of remaining biofilm cells. Further research could explore various forms of surface topographies and different stimuli to enable more effective and reversible changes. In summary, this study reports a new strategy for biofilm control. With further development, it could help reduce medical device-associated infections and biofouling in industrial settings.

On-demand biofilm removal through microscale shape recovery

On-demand biofilm removal through microscale shape recovery

The online version contains supplementary material available at 10.1557/s43577-025-01024-4.

Figures

Schematic of biofilm control by on-demand changes in surface topography. (a) Schematic illustration of the hypothesis – biofilm removal by local changes in surface topography. (b–d) Schematic of the microfabrication process. The programmed shape-memory polymer (SMP) was preheated, then thermally compressed between a patterned PDMS mold with complementary patterns and a flat glass slide to create square-shaped patterns.

Images showing the programmed shape-memory polymer (SMP) surfaces with microscale patterns before and after triggered shape recovery. The tested patterns include spacing (S) and side length (L) of 5 µm, 10 µm, or 50 µm. Scale bar = 10 µm. (Top) Bright-field images. (Bottom) SEM images with titled view of programmed SMP surfaces.

Antibiotic susceptibility of biofilm cells before and after shape recovery. Representative 3D fluorescence images of the *P. aeruginosa* PAO1 biofilms grown on surfaces tested including Control, S5L5, S10L10, and S50L50. Shape recovery was triggered by transferring samples from 0.85 wt% NaCl at room temperature to the same solution at 40℃ and incubated for 10 min. Antibiotic treatment was tested by exposing PAO1 biofilm cells to 32 µg/mL tobramycin (a) or 5 µg/mL ofloxacin (c), for 2 h at room temperature then imaged after Live/Dead staining. (b) and (d) Ratios of SYTO9/PI based on stained bio

Full text

Introduction

Biofilms are sessile communities of bacterial cells embedded in an extracellular matrix consisting of polysaccharides, eDNA, and proteins.^1–3^ Unlikely planktonic counterparts, biofilm cells of the same bacterial strain are up to 1000 times less susceptible to antibiotics due to reduced antibiotic penetration and slow growth of attached cells.^4–6^ The close proximity of biofilm cells also enhances the rate of horizontal gene transfer,^7^ promoting the development of antibiotic-resistant strains.^2^ Biofilm-associated infections thus present a major challenge to the safety of implanted biomedical devices,^8^ such as hip prosthesis,^9^ heart valves,^10^ catheters,^11^ and many others. According to the US National Institutes of Health, about 80% of healthcare-associated infections (HAIs) involve biofilms,^12,13^ emphasizing the needs to engineering better antifouling materials and more effective biofilm control technologies.

A number of strategies have been developed to create antifouling surfaces, including coating with antifouling and/or antimicrobial agents,^14^ using antifouling materials as bulk materials,^15^ biofilm inhibition using patterned chemistry^16^ or responsive materials with bacterial contact triggered dispersion or killing,^17^ and biofilm killing through synergistic effects with conventional antibiotics.^18^ New solutions have shown promising progresses in biofilm control, especially the prevention of biofilm formation. However, removal of established biofilms remains challenging and requires more effective strategies against mature biofilms.

Dynamic changes in surface topography with micron- or nanoscale features have been found effective in removing established biofilms.^16,19^ In the innate immune system of mammalians, bacteria are repelled from the respiratory tract by the coordinated beating of cilia lining the epithelial cells.^20,21^ There are also bioinspired solutions for biofilm control involving mechanically induced motion,^22^ electrically induced deformation,^23,24^ and electromagnetically driven active topography of the material surface.^25^

Previously, we reported that on-demand changes in surface topography of shape-memory polymers (SMPs) can effectively remove biofilms of both bacteria and fungi.^26–28^ SMPs are smart materials that can transition from a temporary shape to a programed permanent shape with triggering by external stimuli, such as temperature, light, electricity, and solvent.^29^ The capabilities to change shape have led to different biomedical applications of SMPs (e.g., wound healing,^30^ drug delivery,^31^ and smart biomaterial scaffold).^32^ Previously, we reported effective biofilm removal by shape recovery of SMP from a uniaxially stretched temporary shape. Upon triggering by heating to 40℃, the material retracted horizontally within 10 min, causing up to three logs reduction of bacterial load on the surface.^26^ The effects are species nonspecific, and detachment sensitizes the cells to antibiotics.^27^ Recent studies of other groups reported biofilm killing^33,34^ and bacteria responsive^35–37^ SMPs, demonstrating the potential of dynamic topography in biofilm control.^38,39^ However, conventional design requires significant changes in the shape of the bulk material, which limits broad applications. This motivated us to develop new strategies that only require local changes in surface topography. In this study, we tested the hypothesis that biofilm removal can be achieved by local changes in surface topography at micron scale alone. To test this hypothesis, we programed acrylate-based SMP with around 6-µm-tall square patterns with varying size and spacing by heat compression, then shape recovered at 40℃ to detach biofilm cells. Escherichia coli and Pseudomonas aeruginosa were used as model species. The results show that local changes in surface topography can detach biofilm cells and increase antibiotic susceptibility. These results further demonstrate the potential of SMPs in biofilm control and can help design better biomaterials to reduce biomedical device-associated infections.

Results

Program SMP for on-demand changes in local topography without altering the bulk shape of the substrate surface

The acrylate-based SMP was successfully synthesized following a previously published protocol;^26^ and differential scanning calorimetry (DSC) results are consistent with reported data.^26,40^ As expected, this SMP has a glass-transition temperature (Tg) ~45℃ and onset temperature (Tg, onset) of 40℃ (Supplementary information Figure S1c). Tg, onset indicates the minimum temperature that allows shape recovery to occur. And Tg implies the temperature that the bulk SMP can be deformed and recovered. Both align with the values reported in literature.^26,40^ All synthesized SMPs were successfully stretched to ~170% (normalized by initial length) followed by immersing in saline buffer or in dry air at 25℃ for 48 h to evaluate shape recovery under wet and dry conditions. With prewetting in 0.85 wt% NaCl aq. solution, complete shape recovery occurred within 10 min when triggered at 40℃. For the same incubation time (10 min), the material recovered to 118% of its initial size when triggered at 37℃. Compared to the prewet conditions, the dry SMPs transferred to a warm buffer without presoaking only showed partial shape recovery (e.g., back to 109%, 122%, and 153% of the original length when triggered in 45℃, 40℃, and 37℃, respectively, in 0.85 wt% NaCl aq. solution). Overall, shape recovery was observed at temperatures close to physiological conditions. Prewetting of programmed SMPs led to faster and more complete shape recovery, which is favorable for biological conditions.

To create micrometer-sized square patterns, we used a mold of polydimethylsiloxane (PDMS) to compress only in the vertical direction (Figure 1). The compression against a plain glass slide transduced a uniform force to create the defined patterns on the SMP substrates, without notable changes in the length, width, and thickness of the base. By immersing the surface in prewarmed 0.85 wt% NaCl aq. solution at 40℃ for 10 min, effective shape recovery was observed with the patterns nearly disappeared.Figure 1. Schematic of biofilm control by on-demand changes in surface topography. (a) Schematic illustration of the hypothesis – biofilm removal by local changes in surface topography. (b–d) Schematic of the microfabrication process. The programmed shape-memory polymer (SMP) was preheated, then thermally compressed between a patterned PDMS mold with complementary patterns and a flat glass slide to create square-shaped patterns.

Validation of shape recovery

Because silicon wafer is rather brittle and we need to apply a compression force to create SMP patterns, we created a PDMS mold with complementary patterns first. The silicon wafer and PDMS mold have features that are 7-µm tall. The heights of all five SMP patterns (S5L5, S5L10, S10L10, S50L50, S100L100) were estimated based on 3D imaging as 5.3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 0.8 µm, 6.3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 0.4 µm, 6.4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 1.0 µm, 6.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 1.3 µm, 6.2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 0.2 µm. The slight variations among patterns are likely due to the elasticity of the polymers that affects pattern formation and peeling. Bright-field images revealed the changes in the surface topography after shape recovery. As expected, shape recovery led to significant reduction of the profile of patterns and surfaces become flatter (Figure 2). In general, the patterns on all programmed SMPs became almost flat after triggered shape recovery at 40℃ for 10 min in 0.85 wt% NaCl aq. solution (Figure 2). The results were corroborated by scanning electron microscopy (SEM) images, which showed titled views of the surfaces and changes in pattern height after shape recovery. The pattern heights are slightly shorter than the mold (7 µm), especially the small patterns (S5L5), indicating that SMP did not fully fill some of the PDMS molds due to elastic behavior of PDMS during heat compression. This also led to slightly round corners of the patterns (Figure 2). The changes in surface topography did not affect the wetness of surfaces (e.g., no significant difference in contact angle between flat control and patterned surfaces) and before and after shape recovery for all the patterns (Figure S6).Figure 2. Images showing the programmed shape-memory polymer (SMP) surfaces with microscale patterns before and after triggered shape recovery. The tested patterns include spacing (S) and side length (L) of 5 µm, 10 µm, or 50 µm. Scale bar = 10 µm. (Top) Bright-field images. (Bottom) SEM images with titled view of programmed SMP surfaces.

Shape recovery triggered removal of P. aeruginosa biofilms

We first evaluated P. aeruginosa biofilm removal from patterned surfaces by shape recovery. As shown in Figure 3, P. aeruginosa biofilms on all tested surfaces (flat Control, S5L5, S5L10, S10L10, S50L50) had a biomass ~27 µm^3^/µm^2^ after 48 h of culturing at room temperature. Upon shape recovery, biomass of the Control remained high (23.9 µm^3^/µm^2^) indicating minimal effects from heat shock used to trigger shape change. In contrast, the biomass of four programmed SMP decreased to 6.1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 1.4, 13.2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 3.6, 7.6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 1.2 and 13.3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 5.9 µm^3^/µm^2^, for S5L5, S5L10, S10L10, S50L50 surface, respectively (Figure 3a). Bright-field and fluorescence images confirmed effective shape recovery with the patterns nearly diminished. This is consistent with sterile samples shown in Figure 2, indicating that shape recovery also occurred in the presence of biofilms. Compared to the samples before shape recovery, dynamic changes in topography caused 71.4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 6.4%, 53.6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 12.5%, and 63.6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 5.9% of biomass reduction for S5L5, S5L10, S10L10 surfaces, respectively. S50L50 surfaces did not show significant biofilm removal upon shape recovery (p > 0.05; Figure 3c). According to the three-dimensional (3D) fluorescence images, the biofilms remained on SMP substrates were in relatively small clusters instead of continuous thick layers before shape recovery.Figure 3P. aeruginosa biofilm removal by shape recovery of protruding square patterns. (a) Bright-field and (b) representative 3D fluorescence images of biofilms. Surfaces tested include Control (static flat surface) and square-shaped patterns with equal side length and spacing of 5 µm (S5L5), 10 µm (S10L10), and 50 µm (S50L50), and squares with side of 10 µm and spacing of 5 µm (S5L10). Shape recovery was triggered by transferring samples from room temperature to 40℃ (both in 0.85 wt% NaCl solution) and incubated for 10 min. (c) Biomass (µm^3^/µm^2^) of biofilms before (orange) and after (blue) shape recovery. Note: Means \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} SE are shown (N \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document} 3). ∗p-value <  0.05, ∗∗p-value < 0.01.

Escherichia coli biofilm removal from programmed SMP surfaces

Because SMPs remove biofilm via physical forces, we expected that is also effective against other bacterial species. To test this, E. coli biofilms were formed on patterned surfaces and tested following the same protocol. The results are shown in Figure 4. E. coli biofilms cultured for 24 h at room temperature had biomass around 1.7 µm^3^/µm^2^. No significant change was observed after 10 min incubation at 40℃. In comparison, the biomass of four programmed SMPs before shape recovery displayed a lower biomass of ~1.1, 1.6, 0.9, 1.3 µm^3^/µm^2^ compared to flat surface, and fewer cells were observed on top of the protruding square surfaces compared to the valleys. This is consistent with our earlier report that square patterns with side length shorter than 20 µm are inhibitory to E. coli biofilm formation.^41^ After shape recovery, the biomass further decreased to 0.54 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 0.04, 0.48 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 0.12, 0.39 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 0.08, and 0.44 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 0.04 µm^3^/µm^2^ on the surfaces with patterns of S5L5, S5L10, S10L10, and S50L50, respectively**.** These correspond to reduction of biomass by 51.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 3.1%, 70.6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 7.1%, 59.2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 8.6%, and 65.2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 3.3%, respectively. The red color in Live/Dead staining images was unprofound for samples both before and after shape recovery, indicating the reduction of biomass was due to biofilm dispersion rather than direct killing of biofilm cells. Overall, these four micro-topographical patterned SMPs also effectively removed E. coli biofilms, demonstrating species nonspecific effects.

Shape recovery increased antibiotic susceptibility of biofilm cells

Because shape recovery caused significant changes to biomass and structure, we speculate that the antibiotic susceptibility of attached cells could also change. To test this, we stained the P. aeruginosa PAO1 biofilms before and after shape recovery with Live/Dead staining kit. The general process was similar to what we described in biofilm imaging. The remaining attached P. aeruginosa PAO1 biofilm cells after shape recovery were exposed to 32 µg/mL tobramycin and treated for 2 h at room temperature in 0.85 wt% NaCl aq. solution.

The images of the Control surface showed healthy biofilms with dominating green color (SYTO9) and no significant change after tobramycin treatment. In contrast, on programmed SMP surfaces, S5L5 and S10L10 biofilms exhibited notable increase in red color (PI staining of cells with damaged membranes) after treatment with tobramycin. For example, the ratio of SYTO9/PI signals were 2.0 and 2.5 for S5L5 and S10L10 after the shape recovery, which are significantly lower than those before shape recovery (6.4 and 5.1, respectively). S50L50 showed a similar trend but weaker dead cell signal compared to other two-patterned SMP with a SYTO9/PI ratio of 4.1 after shape recovery (6.1 before shape recovery). In 3D fluorescence images, the biofilm morphology on all the patterned surfaces before and after shape recovery are comparable to the results in Figure 3b. Similar results were obtained for biofilm treatment with ofloxacin. Specifically, SYTO9/PI ratio of remaining biofilm on S5L5, S10L10, and S50L50 surfaces after shape change dropped to 1.8, 2.6, and 2.6 compared to 8.2, 5.2, and 5.7 before shape recovery, respectively. In summary, the remaining biofilm on micropatterned SMP after shape recovery became more susceptible to tobramycin and ofloxacin, which was not observed for biofilms on flat control surfaces.

Discussion

Our previous study^26^ on acrylate-based SMP demonstrated strong effect of biofilm removal by triggered shape recovery of uniaxially stretched SMP. However, the activity requires significant change in the bulk shape of the substratum, which limits its applications. This motivated us to test the biofilm removal by changes in local topography alone.

The local shape recovery was analyzed via optical imaging. The dimension of the square patterns was evaluated from 3D images. The height of all the square features on SMP was ~6 µm, less than the 7 µm of the depth in the mold due to the resistance during compression against the elastic PDMS. This is consistent across all samples. The dimension of various patterns and shape recovery behavior did not affect the wetness of the surfaces.

It's interesting to note that shape recovery occurred faster in an aqueous environment than dry condition at the same temperature. Contact with a solvent is a factor known to trigger thermoresponsive shape recovery.^42^ Previous studies^43,44^ showed that deformation in water at a temperature below Tg can “soften” PU-based SMP with less than 5% water absorption. And part of the absorbed water works as “bound water” due to the hydrogen bonding between PU-SMP and water molecules, which indirectly affect the transition temperature.^45,46^ In this case, the deformed SMP can have shape recovery in water at a temperature below its Tg. Similarly, Xiao et al.^47^ showed that the solvent is capable of reducing glass-transition temperature by enhancing the chain motility of acrylate-based SMP: The programmed SMPs by uniaxial tension were tested to trigger recovery in dry air and de-ionized water at 25℃ and 30℃. It showed an outstanding recovery in water at both temperatures compared to recovery in dry air. Moreover, both simulation and experimental results from the literature displayed a rubbery response of this SMP with presoaking in water before stretching compared to that stretched in the air, which implies the softening effect of water molecules during shape recovery.^47^ This is advantageous for biomedical applications, which require a physiologically relevant condition, including temperature, pH, and aqueous environment. Our results showed that shape recovery of SMP in saline buffer at 40℃ led to rapid shape recovery from 150 to 105% of the initial size within 5 min.

This study demonstrates the feasibility to achieve biofilm control via dynamic deformation in the z-direction of microscale patterns without changing the bulk shape of SMP. By comparing different square length and interspacing with similar height, it was found that increasing the density of patterns led to stronger detachment of PAO1 biofilms. This is expected because shape recovery occurred mostly vertically in local topographic features. This can break the adhesion of biofilms on the side wall of protruding features more than the those in the valley. Thus, in the areas with denser features, there will be more force generated and also more effects on biofilms in the valley due to close proximity. Consistently, it was observed that the remaining cells after shape recovery were mostly located in the valleys between original patterns. The most condensed patterns (S5L5) achieved ~71% biofilm reduction with remaining cells in small clusters. When the patterns got further apart, the biofilm was less removed (e.g., no significant removal from S50L50 surfaces).

Previously, we reported that static microtopographic patterns affect E. coli formation on PDMS surfaces. Specifically, E. coli forms less biofilm per unit area on top of protruding square patterns when the side length is less than 20 µm.^48^ These inhibitory effects were attributed to the area needed for initial contact with a surface by E. coli flagellar.^48^ This was also observed in this study for E. coli biofilm formation on surfaces with S5L5 and S10L10 patterns before shape recovery, although the effects are less profound, possibly due to the differences in materials (SMP versus PDMS) and actual height (6 µm in the present study versus 10 µm in Reference 41). Further testing of additional patterns, particularly those with taller features and/or different shapes, will help determine whether any synergistic effects exist between static and dynamic factors. Similar to P. aeruginosa results, shape recovery and associated changes in local surface topography also led to 50–65% removal of E. coli biofilms from the patterned surfaces. Together, initial biofilm inhibition and triggered shape recovery led to up to 71% reduction of E. coli biofilm biomass (Figure 4).Figure 4E. coli RP437 biofilm removal by shape recovery of square-shaped patterns. (a) Bright-field and (b) 3D fluorescent images of E. coli biofilms grown on surfaces tested, including Control, S5L5, S5L10, S10L10, and S50L50. Shape recovery was triggered by transferring samples from room temperature 0.85 wt% NaCl solution to the same solution prewarmed at 40℃ and incubated for 10 min. (c) Biomass (µm^3^/µm^2^) quantification of the five shape-memory polymer surfaces before (orange) and after (blue) shape recovery. Note: Means \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} SE are shown (n \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document} 3). ∗p-value < 0.05, ∗∗p-value < 0.01.

In addition to the reduction of biomass, we speculate that biofilm removal could loosen the structure of remaining biofilm and thus increase its permeability and susceptibility to antibiotics. Tobramycin and ofloxacin with reported minimal inhibitory concentrations (MICs) of 3.2 µg/ml and 3.75 µg/ml, respectively, against PAO1 were used in this study.^49,50^ Higer concentrations were used because biofilms are tolerant to antibiotics.^9^ The results of Live/Dead staining revealed that the remaining biofilm cells were sensitized to tobramycin and ofloxacin. The unprogrammed control biofilms exhibited high level antibiotic tolerance after 2 h treatment with tobramycin or ofloxacin. In comparison, shape recovery led to increased susceptibility of the remaining biofilm cells based on SYTO9/PI ratio (Figure 5). Consistently, the remaining biofilms are thinner than the control and dead cells are close to the substratum, confirming that antibiotics can penetrate these cell clusters (Figure** S4**). More attached biofilm cells were effectively killed after treated with 5 µg/mL ofloxacin compared with tobramycin treatment. Besides, with increasing square pattern size, less P. aeruginosa PAO1 biofilms were removed leaving more biofilm clusters with matrix. Thus, stronger SYTO9 signal and less PI signal were observed for S50L50 compared to S5L5 and S10L10. Overall, the Live/Dead assay implies that P. aeruginosa PAO1 biofilm becomes more susceptible to antibiotics after going through the local changes in surface topography, which both detach biofilm cells and sensitize the remaining cells to antibiotics. It is worth noticing that Live/Dead staining is an indirect assessment of viability. Ideally, the Live/Dead staining results should be corroborated with CFU assays. However, it is impractical in the present study because only the local areas near the protruding patterns were going through shape changes, which only account for 25% of the total surface area. The heterogeneity in the sample can potentially overshadow the changes in CFU assay. More studies at single-cell level (e.g., cell sorting [after staining] and plating) could help further evaluate the effects of shape recovery on biofilm detachment.Figure 5. Antibiotic susceptibility of biofilm cells before and after shape recovery. Representative 3D fluorescence images of the P. aeruginosa PAO1 biofilms grown on surfaces tested including Control, S5L5, S10L10, and S50L50. Shape recovery was triggered by transferring samples from 0.85 wt% NaCl at room temperature to the same solution at 40℃ and incubated for 10 min. Antibiotic treatment was tested by exposing PAO1 biofilm cells to 32 µg/mL tobramycin (a) or 5 µg/mL ofloxacin (c), for 2 h at room temperature then imaged after Live/Dead staining. (b) and (d) Ratios of SYTO9/PI based on stained biomass. Note: Means \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} SE are shown (n \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document} 3). ∗p-value < 0.05.

The SMP used in this study can only go through shape change once. Using reversible SMPs could lead to stronger effects of biofilm removal with repeated dynamic changes in surface topography and associated shear force. In a recent study, we reported a  ~94.3% P. aeruginosa biofilm removal from a reversible SMP (rSMP) after three cycles of shape recovery.^28^ Reversible SMPs (rSMPs) with Tg close to body temperature (37℃) and stable physical properties through shape recovery cycles will be good candidates for further study. This is part of our ongoing work. Moreover, different forms of deformation rather than stretching and compassion could be explored. For example, shearing, twisting, or combined deformation could generate stronger forces for more biofilm removal.

Overall, in this proof-of-concept study we demonstrated the possibility to remove biofilms with local changes in surface topography. Further studies are needed to evaluate its potential applications and safety. The transition temperature to trigger shape recovery in this study is 40℃. This is higher than body temperature. One potential application is to remove biofilms from the inner wall of a urinary catheter. Warm saline solution at 40℃ can be injected into the catheter lumen to trigger the shape change and wash away detached biofilms. The catheter wall can be engineered to provide heat insulation and protect urethral tissues. Alternatively, shape recovery can be triggered by other actuators by tailoring the SMP design (e.g., pH, electric field, magnetic field, and light).^51^ The surface topography is created by soft lithography, which can be scaled up because the mold is reusable. In addition, different microfabrication techniques can readily scale up to make larger molds.^2^ Further studies on these topics could provide better solutions to biofilm associated challenges.

Experimental section

Bacterial strains and growth medium

P. aeruginosa PAO1,^25,52^ and E. coli RP437^53^ were grown in Lysogeny Broth (LB) consisting of 10 g/L tryptone, 10 g/L sodium chloride (NaCl), and 5 g/L yeast extract (Thermo Fisher Scientific Inc.) at 37℃ with constant shaking at 200 rpm.

SMP substrate synthesis

The detailed procedure of SMP substratum fabrication has been described previously.^26^ In short, t-butyl acrylate (tBA), poly(ethylene glycol) dimethacrylate (PEGDMA, Mn = 550 Da) at a ratio of 9:1 and photoinitiator 2,2-dimethoxy-2-phenylacetophenone (DMPA) (Sigma-Aldrich, St. Louis, Mo.) were mixed as a precursor solution and injected into poly(dimethylsiloxane) (PDMS) gasket with a 1.2-mm spacer between two glass slides secured by multiple clamps. The polymerization was initiated by exposure to 365 nm UV for 10 min. The system was then annealed at 90℃ for 1 h to complete the polymerization.

Shape recovery dynamics of prewetted SMP

To determine the shape recovery rate of prewet and dry SMP at different temperatures, we prepared SMP specimens with the dimensions of 30 mm (length) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} 15 mm (width) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} 1.2 mm (thickness). All SMPs were preheated to 60℃ for 10 min and then stretched to ~5 cm long. After cooling down to room temperature, the specimens were immersed in 0.85 wt% NaCl aq. solution or in dry air at 25℃ before inducing shape recovery in saline buffer prewarmed at 37℃, 40℃, or 45℃. And the length was monitored in real time to estimate the percentage of shape recovery.

Fabrication of programmed SMP

To obtain the temporary shape, the SMP substrates were cut into 15 mm (length) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} 8 mm (width) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} 1.2 mm (thickness) pieces, placed on a glass slide at 60℃ for 10 min to preheat the polymer, and then compressed between a PDMS mold with recessive square-shaped patterns and the glass slide that held by multiple clips at 60℃ for another 10 min. To prepare the PDMS molds with microtopographic patterns, a silicon wafer with complementary patterns was fabricated via photolithography at Cornell NanoScale Science and Technology Facility (Cornell University, Ithaca, N.Y.). The square-shaped patterns were designed by computer-aided design (CAD) modeling with systematically varied side length (5 µm, 10 µm, 50 µm) and spacing between the adjacent squares (5 µm, 10 µm, 50 µm). The depth of all the features in the mold was 7 µm, which gave the height of PDMS features as 7 µm. The microfabrication process described previously was followed.^54^ Briefly, the pattern design on photomask made by photomask writer was transferred to a silicon wafer, which was spin-coated through the photomask with photoresist after exposed to UV (wavelength range of 405–365 nm). Then the microtopography was created by an etching process. Finally, the perfluorooctyltrichlorosilane (FOTS) were applied to the silicon wafer to obtain hydrophobic finishing. The PDMS then was prepared by mixing of elastomer base and curing agent with a mass ratio of 9:1 (Sylgard 184 Silicone Elastomer Kit, Dow Corning Corp. Midland, Mich.), and poured on the silicon wafer with patterns followed by degassing for 30 min. After curing at 60℃ for 24 h, the PDMS with patterns was polymerized and carefully peeled off from the silicon wafer.

Biofilm removal and biomass quantification

All SMP substrates were sterilized by UV (1 h each side). Two samples of each surface feature were then put in a sterile petri dish containing 30 mL LB medium to compare the biofilms before and after shape recovery. To grow P. aeruginosa PAO1 biofilms, LB was inoculated with an overnight culture to optical density at 600 nm (OD_600_) of 0.05. The culture was incubated at room temperature (25℃) for 48 h. After 48 h of culturing, biofilm covered SMP surfaces were gently washed three times with 0.85 wt% NaCl to remove nonspecifically attached planktonic cells, followed by staining with LIVE/DEAD™ BacLight™ Bacterial Viability Kits (Thermo Fisher Scientific, Waltham, Mass.) for 15 min in dark for imaging. In addition, E. coli was cultured following the same procedure but for 24 h.

Three-dimensional biofilm images were captured using an upright fluorescence microscope (Axio Imager M1, Carl Zeiss Inc., Oberkochen, Germany). The same SMP substrates were then transferred into prewarmed saline buffer to trigger shape recovery at 40℃ for 10 min and imaged again after washing three times with 0.85 wt% NaCl solution to remove detached cells. Three biological replicates were tested for each condition, and six positions were randomly selected from each sample for imaging. The 3D fluorescence images obtained from E. coli biofilms were deconvoluted with the Constrained Iterative method to achieve the best quality using Zen Blue (Carl Zeiss Inc., Berlin, Germany) to remove background noise. Biofilm biomass was quantified using COMSTAT.^55^

Validation of heat triggered shape recovery

All SMP substrates went through the same procedure as described above except that no biofilm was cultured. This allows us to validate heat triggered shape recovery. Specifically, the SMPs were soaked in 0.85 wt% NaCl for 48 h at room temperature before shape recovery. The surface morphologies of all types of SMPs were imaged using an upright microscope (Axio Imager M1, Carl Zeiss Inc., Oberkochen, Germany). And the height of pattern was estimated based on  3D images. To corroborate the results, the samples were imaged using SEM to obtained tilted views. Samples were cut and then sputter-coated with gold for 45 s under 30 mA (Edwards S150A, Edwards, Burgess Hill, UK). SEM images were obtained using JEOL JSM-IT100LA (JEOL Ltd., Tokyo, Japan). Water contact angle was measured via Drop Shape Analysis (ramé-hart Model 250 Goniometer/Tensiometer, ramé-hart Instrument Co., Oslo, Norway) by dropping 5 μL of DI water on the polymer surface.

Antibiotic susceptibility

Three programmed SMP patterns (S5L5, S10L10, and S50L50) were tested along with static flat SMP used as Control. After 48 h of culturing P. aeruginosa PAO1 biofilms on SMP surfaces, each SMP sample after shape recovery (10 min, 40℃ in 0.85 wt% NaCl) was transferred to 1.5 mL 0.85 wt% NaCl with 32 μg/mL tobramycin or 5 μg/mL ofloxacin and treated for 2 h at room temperature (25℃). After rinsing with 0.85 wt% NaCl solution, the SMP samples were stained with Live/Dead staining kit.

Statistical analysis

Error bars in all figures represent the standard error of the mean. All data were analyzed using ANOVA followed by Tukey test. Results with p-value < 0.05 are considered significant. (ns: no significant difference, *p-value ≤ 0.05, **p-value ≤ 0.01, ***p-value ≤ 0.001).

Conclusions

This study demonstrated the feasibility to remove established biofilms by dynamic changes in surface topography alone, while maintaining the bulk shape of the material. A nearly full shape recovery in the z-direction was observed. Increasing the density of topographic features with the same vertical deformation of SMP led to stronger effects. Specifically, the programmed shape recovery removed biofilms of P. aeruginosa and E. coli up to 71.4% and 70.6%, respectively. In addition, the *P. aeruginosa *biofilm cells remaining on patterned SMP surfaces after shape recovery appeared to be more susceptible to tobramycin and ofloxacin.

Supplementary information

Below is the link to the electronic supplementary material.Supplementary file1 (PDF 1074 KB)