Micro-patterned surfaces reduce bacterial colonization and biofilm formation in vitro: Potential for enhancing endotracheal tube designs

Micro-patterned-surface samples and un-patterned (control) surface samples were created as described previously [31]. Briefly, silicon wafer molds were created by transferring the Sharklet pattern design to photoresist-coated silicon wafers through photolithography and the features were deep reactive ion etched to a depth of 3 μm before being cleaned and methylated prior to use. Smooth, un-etched silicon wafers were used to create the un-patterned controls. Dow Corning Silastic T-2 poly-(dimethyl siloxane) elastomer films were cast off the silicon wafer molds by mixing one part by weight curing agent with ten parts by weight resin, degassing under vacuum, pressing to 0.4 mm thick with glass plates over the silicon wafer molds, and curing for 1 hour at 65°C. Each flat 0.4 mm thick silicone film was then punched into 12 mm diameter circular samples and tested as described below.

All clinical isolates analyzed in this study were purchased from ATCC (Table 1). These strains include clinically isolated strains of MRSA, P. aeruginosa, K. pneumoniae, A. baumannii and E. coli. Two additional strains of P. aeruginosa were also assessed to confirm the validity of the results across multiple strains within a species: a second clinical isolate and a laboratory strain. The laboratory P. aeruginosa strain PA14, carrying a ∆bifA mutation, was included in these studies as a comparative strain control due to its genetic modification to overproduce exopolysaccharide and thus form biofilms quickly and consistently [33]. A single colony of each organism, plated on tryptic soy agar (TSA; Criterion), was used to inoculate tryptic soy broth (TSB; Criterion) and grown in a shaking incubator at 37°C and 280 rpm overnight.

Micro-patterned and un-patterned control samples were suctioned to the bottom of a sterile Petri dish around the outer perimeter of the dish using ethanol. Overnight microbial cultures were sub-cultured and grown to early log phase in TSB before centrifuging aliquots and resuspending cell pellets with 1x phosphate buffered saline (PBS; CulGeneX). Samples were immersed in inoculum containing ~10⁷ CFU/mL for incubation time of 1 to 4 hours statically at room temperature depending on the propensity of the species to colonize the surface; the cell density on un-patterned controls was targeted to remain within an average of 2.5 to 6 logs (Table 1). After incubation, the inoculum suspension was decanted and samples were rinsed three times by adding sterile 1xPBS to the dish, rotating for 10 seconds on an orbital shaker set to 80 rpm, and decanting the rinsate. The outer edge of each sample was removed to minimize the variability caused by the un-patterned sidewalls by using a sterile 8 mm biopsy punch (VWR International); the 8 mm samples were then aseptically loaded into a 15 mL conical tube containing 2 mL sterile Dey-Engley broth (Sigma Aldrich). Attached cells were removed and disaggregated from each sample surface by vortexing for 30 seconds, sonicating for 2 minutes in a bath sonicator (Branson 3510) and vortexing an additional 30 seconds [34]. Samples were enumerated by a 10-fold dilution series, plating each dilution onto TSA, and counting the colony forming units (CFU) after sufficient growth at 37°C overnight. Each CFU per sample data point was log₁₀ transformed before statistical analysis. Each experiment was replicated at least three times for each organism; the exact number of experiments conducted for each organism is shown in Table 1. All experiments, except one MRSA experiment, contained four samples per surface type; that MRSA experiment contained three samples per surface type (Table 1). Statistical analysis took place across the three or more experiments for each organism as described below.

Micro-patterned and un-patterned control samples were suctioned to the bottom of a sterile Petri dish around the outer perimeter of the dish using ethanol. For MRSA biofilms: Overnight cultures of MRSA (ATCC700698) were diluted into fresh TSB to a final bacterial concentration of 10⁶ CFU/mL in TSB. Samples were immersed with 20 mL of the bacterial suspension statically at 37°C for four days, with gentle media replenishing each day from the center of the dish. For P. aeruginosa biofilms: Overnight cultures of P. aeruginosa ATCC9027 and P. aeruginosa PA14ΔbifA were diluted into minimal M63 media supplemented with 0.4% arginine (22 mM KH₂PO₄, 40 mM K₂HPO₄, 15 mM (NH₄)₂SO₄, 1 mM MgSO₄), or minimal M63 media supplemented with 0.4% arginine, 2 mg/mL porcine mucin that contains the MucA/B complex found in mucus (Sigma Aldrich), and 400 μg/mL oxacillin (Sigma Aldrich), to a final bacterial concentration of 10⁶ CFU/mL [33, 35‐37]. Samples were immersed with 20 mL of the bacterial suspension statically at 37°C for 24 hours. All dishes were gently rinsed ten times through media exchange by removing 10 mL media from the center of the dish and adding 10 mL sterile 1x PBS. Samples were fixed for 30 minutes in 2.5% gluteraldehyde (Electron Microscopy Sciences) and dehydrated with an ethanol dehydration exchange series (final ethanol concentrations of 25, 50, 75, and 95%). One to three samples were stained with propidium iodide and image stacks were obtained in two to five pre-selected sites per sample by confocal laser scanning microscopy (Zeiss LSM 510 META on Axiovert 200 M). Semi-volumetric analysis was achieved by totaling biofilm area coverage in the x-y plane for each image through the z-stack using ImageJ software after thresholding the image at the brightest frame in the stack [38]. Each biofilm area coverage/image data point was log₁₀ transformed before statistical analysis. All biofilm experiments were completed in triplicate.

A log reduction (LR) per experiment was calculated by subtracting the average log₁₀(micro-pattern data points) from the average log₁₀(un-patterned control data points). After confirming the normality of the log reductions by residual and normal probability plots, the mean log reduction was interpreted as the median percent reduction with the equation: 1-10^{(− LR)}. Statistical significance of the reductions for the colonization and biofilm assays were assessed using a 1-sided t-test of log reductions from at least three experiments per organism. Estimates of the among- and within-experiment variances were assessed using ANOVA of the log transformed cell densities for each un-patterned control sample and micro-patterned sample, with a random effect for experiment. The repeatability standard deviation (SD) of the un-patterned sample, which quantifies the repeatability of the methods based on control variances, is the square root of the sum of the among-experiment variance and the within-experiment variance divided by the number of replicates per experiment [39]. All analyses were performed using the statistical software MiniTab16.

Bacterial colonization on micro-patterned and un-patterned surfaces were evaluated for the top five respiratory pathogens associated with VAP, including clinically isolated strains of MRSA, P. aeruginosa, K. pneumoniae, A. baumannii and E. coli. Two additional strains of P. aeruginosa were also evaluated to confirm the validity of the results across multiple strains within a species: a second clinical isolate and a laboratory, hyper-biofilm forming strain. The micro-patterned surface significantly reduced the colonization of all species and strains compared to un-patterned controls, with median reductions in CFU counts ranging from 95.6% to 99.9%, p < 0.05 when evaluated across multiple experiments (Figure 2). To confirm the consistency of the methods used to obtain these results, the repeatability SD of the log₁₀(CFU/control sample) was calculated across experiments for each organism (Table 1). Except for P. aeruginosa ATCC10197 and K. pneumoniae, all un-patterned repeatability SDs were close to or below 0.50 log, a historical benchmark that indicates highly reproducible colonization on controls across multiple experiments [40]. Even with the increased variability observed for P. aeruginosa ATCC10197 and K. pneumoniae, the micro-patterned surfaces maintained strong statistically significant performance in this assay thereby indicating good repeatability of the log reductions for all of the organisms tested (Figure 2).

Biofilms were grown on micro-patterned and un-patterned surfaces for MRSA and two strains of P. aeruginosa followed by confocal imaging and ImageJ analysis to quantify the bacterial area coverage for each image in the z-stacks. The micro-patterned surface showed a 67% (p = 0.123) median reduction of MRSA biofilm volumes grown over a four day duration in a TSB media when compared to un-patterned control surfaces (Figure 3). Similar observations were seen for the hyper-biofilm forming strain P. aeruginosa PA14ΔbifA, where a significant 52% median reduction in volume (p = 0.05) was observed (Figure 3). Importantly, mature biofilms [17] were obtained on the un-patterned control surfaces for both the MRSA and P. aeruginosa PA14ΔbifA strains, whereas bacterial colonization on the micro-patterned surfaces was more dispersed (Figure 3). When quantifying the resemblance of the biofilm volume measurements on un-patterned controls, the repeatability SDs of the log-transformed biofilm coverage data was 0.081 for the PA14ΔbifA experiments and 0.681 for the MRSA experiments. While the level of acceptable variability of image analysis data has not been established, the small SDs for PA14ΔbifA indicates high consistency in biofilm formation on the un-patterned controls. The P. aeruginosa ATCC 9027 strain did not form a sufficient quantity of biofilm under these non-clinical conditions to accurately quantify biofilm volume coverage, as most image coordinates had no depth from which to obtain image stacks (Figure 3).

The P. aeruginosa clinical isolate ATCC 9027 did not form biofilms in the simplified minimal medium (Figure 3), so additional studies on this strain were conducted in a medium that would both encourage biofilm formation and simulate the clinical environment of the lungs. Thus, a mucin-rich minimal medium supplemented with oxacillin was used to promote biofilm and provide the mucin glycoproteins that make up the main component of the mucus present in the lungs [22‐24, 35, 36]. This medium change resulted in enhanced biofilm formation of this clinical isolate on the un-patterned control surfaces, whereas the micro-patterned surfaces maintained an inhibitory effect, with a 58% reduction compared to controls (p = 0.009) (Figure 4). The repeatability SD for biofilms on controls across three experiments was low at 0.11, suggesting that the methodology used in these assays obtains repeatable biofilm growth and coverage.