Combining Microscopy Assays of Bacteria-Surface Interactions To Better Evaluate Antimicrobial Polymer Coatings

ABSTRACT

Validation of the antimicrobial performance of contact-killing polymer surfaces through the experimental determination of bacterial adhesion or viability, is essential for their targeted development and application. However, there is not yet a consensus on the single most appropriate evaluation method or procedure. Combining and benchmarking previously reported assays could reduce the significant variation and misinterpretation of efficacy data obtained from different methods. In this work, we systematically investigated the response of bacterial cells to antiadhesive and antiseptic polymer coatings by combining (i) bulk solution-based, (ii) thin-film spacer-based, and (iii) direct-contact assays. In addition, we evaluated the studied assays using a five-point scoring framework that highlights key areas for improvement. Our data suggest that combined microscopy assays provide a more comprehensive representation of antimicrobial performance, thereby helping to identify effective types of antibacterial polymer coatings.

IMPORTANCE We present and evaluate a combination of methods for validating the efficacy of antimicrobial surfaces. Antimicrobial surfaces/coatings based on contact-killing components can be instrumental to functionalize a wide range of products. However, there is not yet a consensus on the single, most appropriate method to evaluate their performance. By combining three microscopy methods, we were able to discern contact-killing effects at the single-cell level that were not detectable by conventional bulk microbiological analyses. The developed approach is considered advantageous for the future targeted development of robust and sustainable antimicrobial surfaces.

INTRODUCTION

Microbes are integral to the survival of higher species, but controlling the growth of pathogenic species is key to preventing bacterial infections and a global health crisis-exacerbated by increasing antibiotic resistance (1, 2). Bacterial adhesion is the primary stage of bacterial infection, whereby planktonic bacteria colonize surfaces before invading the host or forming biofilms (3–5). Functional surfaces (e.g., selectively adhesive, antiadhesive, or antimicrobial) are designed to leverage a versatility of responses in the interaction between animate and inanimate objects.

To control bacterial infections that may be associated with biofilm formation on biomedical devices, water treatment equipment, or food packaging, among other applications, functional polymer coatings become a powerful and versatile tool compared to conventional (bulk) disinfection methods (6–8) by preventing the initial steps of bacterial adhesion (9, 10). This surface functionalization is quickly emerging as a leading frontier in confronting bacterium-related infections (11–13). Strategies for the design of antiadhesive approaches are often based on antiadhesive polymer systems (e.g., polyethylene glycol [PEG] brushes) (14, 15). Antiadhesive polymer coatings can additionally be equipped with a bioactive compound, for example, conjugated with metal nanoparticles, antiseptics, or antibiotics (16–18). Of the various approaches proposed, contact-killing surfaces decorated with biocides that require direct bacterium-surface contact to cause membrane failure and subsequent cell lysis remain one of the safest approaches for a number of applications (6, 19–22). Among these, surface-bound quaternary ammonium compounds (QACs) have been successfully demonstrated to confer contact-killing capabilities (19, 20, 23–26), even in the context of point-of-use water treatment. However, a minimum biocide density of 10¹⁴ to 10¹⁵ positive charges/cm² is required to induce an electrostatic potential for optimal activity (23, 27, 28). Hexetidine is an alternative cationic compound frequently used as a disinfectant in mouthwash (29–31), which inhibits the growth of bacterial pathogens, fungi (29, 32), and yeast (33). The combination of cationic ammonium and amine groups makes hexetidine a promising alternative to QACs; surface tethering can be achieved with minimal manipulation of the biocide. An alternative strategy to develop antiadhesive interfaces is based on biomimicry, which adopts surface topography from various species that prevents microbial adhesion (34, 35).

Standardized performance tests of antimicrobial polymer coatings should confirm both tight interfacial immobilization of the bioactive compound and antimicrobial efficacy. To date, the appropriate evaluation of antimicrobial activity is controversial (36, 37), which is a major obstacle to technology transfer, especially for contact-killing surfaces (19, 38, 39). The most accepted tests rely on the laborious process of colony counting: the solution-based ASTM E2149-13a test (40); the thin-film Japanese JIS Z 2801 assay, which was upgraded to ISO 22196 in 2011 (41, 42); Petrifilm assays (3M, St. Paul, MN, USA); and various modifications thereof. A number of reviews have challenged the variability of the results due to the protocols used (36, 38, 39, 43, 44) or diverse physiological conditions applied (45, 46), which makes it difficult to reproduce and reliably compare antimicrobial performances on surfaces (37). van de Lagemaat et al. conducted a comparison of the above-mentioned methods to determine contact killing on cationic surfaces (39). In a recent review (43), the evaluation methods were categorized according to their suitability in terms of antimicrobial mode of action, but no single standard test could meet these distinctive criteria without modifications. The results of these critical reviews suggest that the thin-film Petrifilm and JIS Z 2801/ISO 22196 assays are appropriate methods for evaluating bacterial contact-killing surfaces but only when used carefully in conjunction with a Hemmhof (zone-of-inhibition [ZOI]) agar assay (47) to exclude death from leachable biocides. However, Bruenke and coworkers (48) argued that nonleachable compounds often fail this agar test due to the low loading capacity of covalently bound biocide monolayers. To improve the suitability of these tests, a modified fluorescence-based protocol for a live- and dead-cell assay was introduced, in which the bactericidal effect of two separated surfaces is monitored (spacer experiment), to determine whether the bactericidal effect occurs exclusively on the active surface (38).

Despite modifications of the above-described assays, the surface efficacy readouts are mainly limited to reporting on reduced bacterial adhesion or cell viability. However, these results are heavily biased toward bacterial culturability and mixed readouts from live/dead fluorescence staining (49). Given the limitations of both solution-based and thin-film assays (highlighted in Fig. 1), alternative methods have been proposed to more adequately determine bacterium-surface interactions. Microscopy studies of single-cell growth kinetics are commonly used in microbial growth studies (50–52) but have, to the-best-of-our-knowledge, very rarely been applied to evaluate functional polymer surfaces (53). Insights into the proliferation rate and daughter cell release significantly influence infection development (48). To this end, we propose to complement bulk population studies with single-cell growth analyses zooming in on microbial interfacial interactions when in direct contact with the active surfaces. Downscaled single-cell experiments not only provide a pristine view of growth inhibition but also allow real-time monitoring of interfacial interactions between individual bacterial cells and active surfaces, for example, when studying antibiotic-related cell death by cell lysis (54). At the same time, due to the “artificial” microenvironment used, these approaches introduce bias results; effects attributable to cell-cell interactions, nutrient competition, as well as Brownian motion and motility are attenuated, requiring caution when interpreting the obtained data (55). Here, we suggest that the inclusion of single-cell experiments provides a more comprehensive description of microbial interfacial interactions to complement predominantly bulk-driven population studies (55, 56).

FIG 1.

Schematic summary and comparison of the three microscopy-based assays combined in this study to determine the antimicrobial performance of functionalized polymer surfaces.

With this background in mind, the main objective of this study was to investigate the applicability of three different assays (bulk, thin film, and direct contact) with their advantages and disadvantages (Fig. 1) to adequately evaluate the behavior of Escherichia coli on functional polymers equipped with antiadhesive and antiseptic moieties, namely, hexetidine and quaternary ammonium compounds, using a recently introduced adsorptive polymer surface functionalization technology (57). The coating approach relies on customized polystyrene (PS) maleic anhydride copolymers, which are referred to as “anchor polymers” (APs) due to their robust surface association ability from aqueous solution. The PEGylation and decoration of the adsorbable polymers with bioactive compounds lead to a great versatility of the coating technology, which is suitable for a wide range of applications (57). The work presented here is part of a larger ongoing study exploring the prospects of bifunctional antimicrobial coatings for point-of-use water treatment. In this work, we used the coating principle as an example for the antiseptic functionalization of polyethylene (PE) surfaces, which are the reference polymer material for the transport and storage of drinking water systems (58). Our developed combination of assays revealed potentially useful effects of functional coatings on interfacial bacteria that would be difficult to detect using conventional bulk microbiological assays. In addition, we apply a technical evaluation framework to compare each assay against five key criteria and provide a useful benchmarking tool for future work.

RESULTS

Solution-based assay.

The antimicrobial activity of functionalized polymer surfaces was first assessed using the solution-based assay combined with live- and dead-cell staining (Fig. 1). The APs selected for this study provide a backbone architecture sufficient for tight surface association on polyethylene, with a backbone-tethered active compound to confer both antiadhesive and bactericidal properties (57). The AP nomenclature MW_bb-MA%-P_EG/S-MW_PEG displays the molecular weight of the backbone (MW_bb in kilodaltons), the percentage of maleic acid groups (MA%), the number of PEG units per number of styrene units (P_EG/S), and the molecular weight of the PEG chain (MW_PEG) (see Table S1 in the supplemental material for the APs applied in this work) (57). Here, we used E. coli strain W3110 as the model contaminant bacterial species as it is the most common predictor of potential risks in drinking water (59, 60). E. coli W3110 was studied in diluted M9* medium (see Materials and Methods for more details) and under bulk conditions with a high volume-to-surface ratio. The surfaces, with the exception of the 70-50-11-0.5 (see Table S1 for code details) coating, showed reasonably comparable surface attachment (Fig. 2A). 70-50-11-0.5 is an antiadhesive polymer that exhibited significantly reduced adhesion behavior, with approximately 60% less surface coverage, which is expected due to the polyethylene glycol component that reduces surface interactions. We found that the antiseptic surface (70-50-11-0.5-Hex) demonstrated the highest dead/live ratio, with approximately 80% dead cells (Fig. 1B), which makes the 70-50-11-0.5-Hex coating approximately 1.5 times more effective in killing bacterial cells than the covalently surface-bound quaternary ammonium compound (QAC control), with a dead/live ratio of approximately 50% (Fig. 2B). The control surfaces (glass and nonfunctionalized polymer) showed similar dead/live ratios of approximately 20 to 30% (Fig. 2B).

FIG 2.

Solution-based assay: impacts on bacterial viability of E. coli W3110 of glass control surfaces, antiadhesive surfaces (70-50-11-0.5), and antiseptic surfaces functionalized with hexetidine (70-50-11-0.5-Hex) or quaternary ammonium compounds (QAC control) on polyethylene-coated glass. The data depict in situ endpoint live/dead staining, after incubation for 4 h in modified M9* medium (M9* minimal salts supplemented with glucose and casein hydrolysate for final concentrations of 0.004% glucose and 0.02% casein hydrolysate) at 30°C with constant shaking (60 rpm). (A) Percent bacterial surface area coverage of live and dead fluorescence. (B) Relative percentages of live and dead bacterial cells (calculated as bacterial viability) of adhered E. coli W3110 cells. Note that the death of bacteria in both controls (glass and 70-50-11-0.5) may be caused by lowered nutrient levels in the diluted medium (error bars denote standard deviations) (ns, not significant; *, significance at a P value of <0.02 for total surface coverage; ****, significance at a P value of <0.001 for bacterial viability [by one-way analysis of variance {ANOVA}]) (n = 3 independent biological replicates).

Spacer thin-film assay.

The general spacer thin-film experiment is based on the standardized JIS Z 2801 or ISO 22196 thin-film test (41, 42). Here, we used the modified version for immobilized biocides that incorporates fluorescence, as described previously in a modified live/dead protocol, to monitor biocidal activity on two separated surfaces (38). In our approach, we make use of a physical spacer (120 μm in height) that separates three zones of interest (Fig. 1): the functionalized surface at the bottom (active surface area), an unmodified glass slide at the top (glass control), and a volume fraction of approximately 8 μL of a prestained bacterial solution in the middle. Bacterial viability in terms of the dead/live ratio of cell coverage was continuously monitored in all 3 zones for up to 3 h with a fluorescence microscope. However, due to the dynamic nature and time constraints in the volume phase, we focused our analysis on the direct bacterium-surface interactions only.

In our assay, we found bactericidal effects for the 70-50-11-0.5-Hex surface similar to those observed in the bulk assays but with a dead/live ratio of approximately 90% (Fig. 3A) in 3 h. We also observed that the overall percent coverage of bacteria (Fig. S1C) on the substrate surface was approximately 4 times higher than that measured in the bulk assays, which had approximately 200 times the volume phase under dynamic (shaking) conditions. In the spacer experiment, we did not observe a significant bactericidal effect for QAC surfaces compared to the control surface (glass or 70-50-11-0.5). For all surfaces, we observed an initial spike in dead cells on the active surface (Fig. 3A), before a steady decline after about 40 min. No clear relation in terms of antiadhesive behavior was observed for all surfaces. Nonetheless, we monitored and evaluated the dead-cell coverage (percent fluorescence intensity per image view) of E. coli in the volume fraction at a set z-plane (Fig. S1B). Here, the dead-cell counts close to the 70-50-11-0.5-Hex surface spiked to 1% in the first 90 min, but otherwise, dead-cell coverage remained low for all test scenarios (approximately 0.1%), suggesting that dead cells are rarely found in the volume phase and that the killing is limited to the surfaces, and lysis of dead cells occurs at the surface. Live-cell counts fluctuated, with cells continuously moving in and out of the frame, presumably due to cell motility, sedimentation, growth, and adhesion to the two surfaces. Hence, no conclusive observations could be made from the “center zone” using the current experimental setup.

FIG 3.

Spacer experiments: tracking bacterial viability (dead/live ratio) of E. coli W3110 over time on 70-50-11-0.5-Hex and quaternary ammonium compound (QAC control) surfaces, on polyethylene (PE) and functionalized glass, respectively, in comparison to the polymer and glass controls. Results depict in situ live- and dead-cell staining, taken as a time-lapse image every 8 min up to 4 h in modified M9* medium (M9* minimal salts supplemented with glucose and casein hydrolysate for final concentrations of 0.004% glucose and 0.02% casein hydrolysate) at 30°C. The dead/live ratio per image based on the average fluorescence intensity is depicted for the bottom substrate (functionalized surface) (A) and the top surface (glass) (nonfunctionalized surface) (B) (shaded areas denote standard deviations) (n = 3 independent biological replicates).

Direct-contact single-cell growth kinetics assay.

To further expand on the information extracted from the assessments for functionalized bacterium-surface interactions, we applied an experimental method that is commonly used in bacterial growth studies: single-cell growth studies (51, 52, 61). Contrary to the above-described solution- and spacer-based assays, the assay design forces bacterial cells into direct contact by sandwiching the cells between a medium-infused agar pad and the active surfaces. Direct contact should facilitate bactericidal, antimicrobial, or antiseptic efficacy. This design enables us to in situ monitor the effect of surface modifications on bacterial growth dynamics, which results in normal exponential growth (control surfaces), growth inhibition (bacteriostatic), or, in some cases, cellular lysis/killing (bactericidal).

The bacterial cells are sandwiched between nutrient agar pads, and the test surfaces are imaged in phase contrast every 5 min to record single-cell-length tracks (Fig. 4A; for details, see Materials and Methods). We calculated the increase in the relative bacterial cell length (ΔL/L₀) for individual cells (Fig. 4B) and the absolute growth rates (Fig. 4C). Growth was unaffected by the presence of antiadhesive polymers compared to the controls on pure glass surfaces. On all surfaces, we observed bacterial cell proliferation (Fig. 4 and Fig. S2). However, we found that there was a wide distribution of cell growth rates for E. coli on the functional surfaces with active compounds. Despite previously identified bactericidal efficacies from bulk studies, we see individual growth on 70-50-11-0.5-Hex and on QAC control surfaces. The growth distribution for 70-50-11-0.5-Hex and QAC control surfaces ranges from reduced to low and even negative growth rates, which indicates that cell growth is inhibited and/or that some cells are actively lysing, while becoming shorter first (compare individual tracks in Fig. 4A), which is a common response prior to lytic events (54). Despite the lower bactericidal effects observed in both the bulk and reduced-volume spacer experiments, the growth rate of cells on the QAC control was significantly lower than that of the control and at a rate 0.03 μm/min lower than that of the most bactericidal surface, 70-50-11-0.5-Hex. The differences between the different methods are a direct indication of the methodology. Here, we focus on the effects of surface chemistry on direct contact with individual cells within a much shorter time frame, while in bulk and spacer experiments, the contact duration is at least twice that of our observations at the single-cell level.

FIG 4.

Growth analysis of single cells using the direct-contact assay. (A) Single-cell growth curves in absolute length increases for E. coli W3110 on glass (brown) (n = 185 cells), 70-50-11-0.5 (gray) (n = 62 cells), 70-50-11-0.5-Hex (blue) (n = 166 cells), and the QAC control (green) (n = 84 cells). (B) Relative length increases for E. coli W3110 on all substrates derived from panel A (means ± SD). (C) Growth rates for E. coli W3110 on all substrates in micrometers per minute (n = 3 independent biological replicates) (n.s., not significant; ****, P < 0.0001 [by Student’s t test]).

DISCUSSION

Solution-based assays.

Immersive inoculation is one of the most common antimicrobial assessment methods (42). The most widely used method, ASTM E2149, assumes that the reduction in CFU is due to the mass transport of bacteria to the active surface and that direct contact with the bactericidal surface causes the reduction in cell numbers (28, 37, 49, 62). Thereafter, dead cells must be released into the bulk phase for surface killing to continue. Over time, cell debris would form a conditioning film that gradually blocks the active surface compounds. However, this method does not preclude the elution of biocides from the active surface, which may also contribute to solution killing. On the other hand, over a longer incubation period, nutrient-poor growth conditions may also reduce the CFU count. Taken together, these facts make this test method inconclusive for evaluating the contact-killing efficacy of active surfaces. Variabilities in CFU readouts due to (i) inadequate cell desorption driven by cell surface adhesion forces (63, 64), (ii) the reversible “dead” state of viable but not culturable bacteria (65, 66), (iii) the high susceptibility of planktonic bacteria to biocides (67, 68), (iv) the inoculum limit of surfaces (27), and (v) the dynamics of biocide leaching and subsequent concentration gradients (16), limited by the high volume-to-surface-area ratio, are among the main drawbacks highlighted in the literature (16, 36, 43) for such solution-based assays. In addition, dead cells can be adsorbed or desorbed from surfaces at different rates, which can confound the assay results and make it difficult to detect surface-induced killing.

We analyzed and evaluated the bactericidal efficacy of the live/dead, immersive method used here, adapted from methods described previously by Cavallaro et al. (27). Adaptation of the method allows a better display of the contact-killing potential in lieu of leachable biocides while simultaneously describing antiadhesive behavior. However, similar to the standardized ASTM E2149 assay, the dynamic conditions and the high volume/inoculum-to-surface-area ratio impact mass transport as well as successful cell-surface collisions required for contact killing. Substrate modifications result in differing surface chemistries; i.e., in our study, the hydrophobic 70-50-11-0.5-Hex or the hydrophilic 70-50-11-0.5 surfaces also affect the adhesion behavior toward the surface, which could decrease surface contact. Overall, the suspension methods are not always ideal or realistic scenarios for various applications to assess contact-killing surfaces.

Thin-film assays.

Assays with lower volume-to-surface ratios increase the probability of direct contact between planktonic bacteria and the active surfaces. Thin-film assays not only facilitate the buildup of higher biocide concentrations in the case of biocide-eluting surfaces but also are an ideal setup for contact-killing surfaces. Here, the JIS Z 2801 assay or its upgraded version, the ISO 22961 assay, is the standard assay for antimicrobial plastic surfaces. A thin film of bacteria is sandwiched between the active surface and a plastic film and incubated for 24 h, after which the cells are extracted from the surface by sonication and enumerated by CFU counts on agar plates. A more direct method to circumvent the washing/sonication step replaces the top plastic film with Petrifilm, a transparent foil that is coated with agar infused with a tetrazolium chloride stain. Both approaches must be used in conjunction with an agar zone-of-inhibition (ZOI) assay to confirm that the active biocide is either leachable or tightly immobilized. In our case, no zone of inhibition was observed for either antiseptic surface (see Fig. S3 in the supplemental material), which may indicate tight immobilization of bioactive compounds. However, contact-killing surfaces may fail this test (48). Typically, the biocide loading capacity of immobilized surfaces is often lower than the MIC required to see an effect.

To avoid the above-mentioned challenges with CFU counts, we modified the standard JIS Z 2801/ISO 2296 assay to a time-lapse spacer experiment using in situ fluorescence microscopy. This method was previously described as a 3-in-1 assay that enables the quantification of (i) the potential immobilization of a biocide agent on a surface, (ii) the death rate, and (iii) the adsorption/desorption behavior of dead cells (49). It has the benefit of providing additional in-depth information about the bactericidal action on the active substrate and the nonactive control surface in juxtaposition, and, if carefully handled, it can be used to assess the dead/live ratio of bacteria in the volume fraction. Instead of plating and CFU counting on agar plates, we again adopted live/dead staining to continuously monitor bacterial behavior. In our adapted protocol, we assemble the assay by using imaging spacers (120 μm) to avoid bacterial adhesion to the spacer glass beads, as originally described by Green et al. (49).

Our results imply that the effect of the reduced volume-to-surface-area ratio successfully demonstrated an increase in bactericidal efficacy compared to the bulk assay. Using live/dead staining, we showed that the 70-50-11-0.5-Hex surface was highly bactericidal. However, when we performed testing as suggested for the standardized JIS Z 2801/ISO 2296 assay, we found only a 0.8-log reduction in the CFU count, with large variability between independent experiments (see Fig. S4A in the supplemental material). We conclude, based on the different extraction methods used (Fig. S4B), that the insufficient and inconsistent removal of bacteria from the active surface (or mixed with the volume phase) is a major source of variability in readouts (37, 46). The QAC surface experienced a 40-min window where the dead/live ratio increased to 50% before declining to a level comparable to that for the nonfunctionalized control after 90 min. Asri et al. demonstrated the irreversible membrane binding of bacterial cells to QAC surfaces as a mode of action (23). In the absence of stirring, cell debris could block active sites and lead to a decline in activity over time (27, 69). The design of the spacer assay allows the time-dependent tracking of not only the adhesion and desorption of bacteria on the active surface, as well as the control surface, but also the movement of bacteria into the volume fraction. However, veering on the side of caution, we considered only bacterial behavior on the separated surfaces, as three-dimensional (3D) tracking of individual cells in all 3 regions would require a faster imaging process than our microscopy technique is capable of.

In our hands, prestaining, particularly concerning propidium iodide (PI) toxicity, had little effect on bacterial viability overall because the dead-cell ratio remained relatively constant in both regions for the glass control. However, in both the solution-based and spacer experiments used here, quantitative errors may arise from the unequal or dual staining of DNA (70, 71) and STYO9 bleaching (70) as well as cell debris, which was difficult to distinguish during our analysis of fluorescence images. Added to this, previous studies have reported that red-stained cells, which were presumably dead, were still culturable (72, 73). Generally, the evaluation of death, in the absence of visible cell lysis, remains a major challenge for microbiologists (74, 75).

Single-cell, direct-contact assays.

To better understand how the volume phase influences the bactericidal and bacteriostatic behavior on surfaces, we attempted to establish an assay that enables us to decouple volume behavior from direct surface contact. We sandwiched bacteria directly between the active surface and a medium-containing agar pad to measure the effect of the active surface at the single-cell level. The readouts in the two volume-phase methods are live- and dead-cell staining, but since there are large volume fractions involved, it becomes complex to disentangle the origin of cell death in the two volume assays. At the single-cell level, we were able to track individual bacterial cells and either their growth, growth inhibition, or even lysis (killing). Tracking single-cell growth showed that there is a wide spectrum rather than a universal response to the active surface for individual cells. These responses range from no response to bacteriostatic and bactericidal activity. Furthermore, a more in-depth analysis will allow the extraction of additional information about bacterial metabolic states and bactericidal activity or death, which was not the focus of this study. For active functionalized surface coatings that we presumed to be ineffective in the solution-based and spacer assays, we learned that there is still a wide range of potential interactions between bacteria and the surface. Even though the surface is not bactericidal as a whole, we found that the observed short-term inhibition of growth and even lysis of cells may be crucial for various biomedical applications to increase the chances for the immune system to address a response and wipe out pathogenic strains. Growth inhibition may be advantageous for applications such as short-term catheter insertion procedures.

Evaluating the studied antimicrobial surface assays.

In the discussion of the criteria that make up the “ideal” test, there is a wide range of applicable assays to choose from, depending, of course, on the conditions of use (36). In this work, we have emphasized that combining assays can mitigate the substantial variations and misinterpretation of efficacy data obtained by different methods to compensate for the shortcomings of a single assay. Future work will benefit from an evaluation framework that assesses and guides the systematic combination and selection of existing tests. Below, such an evaluation framework is developed for the studied assays to assess the efficacy of antibacterial surfaces based on technology assessment. Technology assessment is an analytical practice for the early-stage identification of potential impacts of technological change with the goal of gaining public and/or political opinion on societal aspects of science and technology (76, 77). It is a decision-making tool that can take many forms and is conducted by (or for) organizations or as a medical, life cycle, or sustainability assessment (78). Here, we adopt a technical assessment lens to build an evaluation tool that considers five key features/criteria (in our opinion) of an ideal antimicrobial assay: (i) flexibility, (ii) ease of use, (iii) reliability, (iv) informativeness, and (v) scalability/high-throughput potential. The descriptors were carefully chosen (based on our experience and that of reviews) to represent useful and general characteristics for the assays (38, 43). Flexibility refers to the types of substrates that can be analyzed in terms of geometry, opacity, and reflectance. The ease-of-use parameter evaluates complexity: the number of process steps, material consumption, and time required for the assay. Reliability is directly extracted from the variability of the results (standard deviations [SD]) and the human error that occurs in indirect analysis or enumeration (degrees of freedom). Informativeness refers to the specific aspects of antimicrobial performance that an assay can measure, e.g., immobilized or leachable biocides, measures of viability, adhesion behavior, and/or growth inhibition. Finally, high-throughput potential evaluates scalability, the potential for certain processes to be automated, as well as a simple cost-benefit analysis (cost [euros] per sample). Our coating platform gives ample room to explore precision and high-throughput antimicrobial surface design, for which the fifth criterion becomes important. We thoroughly evaluated and scored each criterion for all described assays and plotted the results in radar graphs (Fig. 5; for detailed scores, see Tables S2 to S5 in the supplemental material). We can deduce that the flexibility of the spacer (2), solution-based (1), and direct-contact assays is somewhat limited to the morphology of the sample (Table S3): only flat transparent substrates are suitable for these assessments. Under this criterion, the JIZ assay and the ZOI tests are more “flexible,” allowing a wider range of geometric shapes to be tested. However, these two assays rely on indirect CFU enumeration, which is not the most reliable and informative assessment. Although an ideal assay would score “5” on each criterion, we think that the direct-contact assay holds great untapped potential for the community, especially given the tremendous wealth of information at the single-cell level (information for growth, stalling growth, morphology, and lysis), which is unprecedented compared to other assays. This tool can be leveraged to improve current testing methods or to systematically combine assays in order to build an ideal setup or test workflow for intended applications.

FIG 5.

Benchmarking and technical assessment tool for solution-based (A), spacer (B), direct-contact (C), JIS Z 2801 (JIZ) (D), zone-of-inhibition (Zone) (E), and idealized optimum (F) antimicrobial assays based on five main criteria: (i) flexibility, (ii) ease of use, (iii) reliability, (iv) high-throughput (ht) potential, and (v) informativeness (81). A more detailed description of the evaluation process can be found in Tables S2 to S5 in the supplemental material.

Conclusion.

The combination and benchmarking of different microscopy assays for the evaluation of antimicrobial polymer coatings have shown that the limitations of using a single method or standardized assay can be avoided. Solution-based methods with high volume-to-surface-area ratios were demonstrated to be poorly suited for evaluating contact-killing surfaces. Instead, the determination of live/dead cells directly on the surface more sensitively indicated the bactericidal capacity of functionalized surfaces. Moreover, forcing bacterial cells to contact antimicrobial surfaces by reducing the volume fraction in thin-film assays proved to be a suitable method to demonstrate contact killing. We anticipate that our approach of direct-contact experiments at the single-cell level will help identify useful functional features of antimicrobial coatings that are difficult to assess using conventional CFU analyses from volume assays.

MATERIALS AND METHODS

Test surfaces.

The antiadhesive anchor polymer (70-50-11-0.5) is a variant of an amphiphilic copolymer toolbox described previously (57). Modified hexetidine (hexetidine, mix of stereoisomers; Sigma-Aldrich, Germany) was tethered to this anchor polymer to form an antiseptic version of an easy-to-apply, antiseptic polymer surface, 70-50-11-0.5-Hex. A quaternary ammonium compound (QAC) in the form of glycidyl-trialkyl-ammonium chloride (Sigma-Aldrich, Germany) was covalently bound to 3-aminopropyl-triethoxysilane (Sigma-Aldrich, Germany)-functionalized glass surfaces, using a method previously proven to demonstrate contact killing (27) without leaching (QAC control). All samples were based on glass coverslips of 0.14 mm and 24 by 24 mm (Marienfeld, Carl Roth, Germany).

Bacterial strains, culture media, and growth conditions.

Escherichia coli strain W3110 was streaked from glycerol stocks (−80°C) onto Luria-Bertani (LB) (Sigma-Aldrich, Germany) culture plates, incubated at 37°C overnight, and stored at 4°C. A single colony was inoculated overnight in 2 mL of M9* minimal medium (modified and supplemented M9 medium [M9 minimal salts {catalog no. M6030; Sigma-Aldrich, Germany} supplemented with 2% casein hydrolysate broth {Sigma-Aldrich, Germany} and 0.4% glucose {20% glucose; Teknova, VWR Germany}]) at 30°C at 220 rpm overnight. The optical density (OD) was determined on a spectrophotometer (Eppendorf BioPhotometer). Prior to each experiment, the culture of bacterial cells grown overnight was diluted 1:200 in fresh M9* minimal medium and grown to mid-exponential phase (10⁸ CFU/mL; OD at 600 nm [OD₆₀₀] of ∼0.3). Depending on the experimental setup, the exponential-phase cells were either used directly or otherwise suspended in the test medium. For the latter, bacterial cells were centrifuged at 3,000 × g for 3 min (Universal 320 R; Hettich, Germany), the supernatant was discarded, and the cells were resuspended in modified M9* medium (we supplemented M9* minimal salts to prepare modified M9* medium with glucose and casein hydrolysate for final concentrations of 0.004% glucose and 0.02% casein hydrolysate) for two cycles. The starting numbers were normalized by resuspension in the test medium, equivalent to approximately 10⁶ CFU/mL, corresponding to an OD of ∼0.04.

Detection of leachable compounds.

To rule out potential effects of leached biocides and confirm contact killing as suggested previously by van de Lagemaat et al. (39), thoroughly rinsed (3 times in MilliQ water) functionalized surfaces were put facedown on inoculated (LB) agar plates (10⁸ CFU/mL) and incubated for 24 h at 30°C. A visual account of the zone of inhibition surrounding the substrates was assessed the next day.

Solution-based assay (live and dead cells).

Live/dead staining gives an indication of possible membrane damage by a bioactive compound. In this assay, E. coli W3110 was used as a model organism. Here, we modified a bulk-solution-based protocol described previously by Cavallaro et al. (27) to evaluate the antimicrobial efficacy of QAC-functionalized glass. In our approach, an inoculum regrown overnight to the exponential phase was rinsed and diluted in modified M9* medium (we supplemented M9* minimal salts to prepare modified M9* medium with glucose and casein hydrolysate for final concentrations of 0.004% glucose and 0.02% casein hydrolysate) to 10⁶ CFU/mL (OD of ∼0.04) (Eppendorf BioPhotometer). Within a 6-well plate (polystyrene [PS]) (Costar, catalog no. 3736; Corning), 2 mL of the test inoculum was incubated with the 24-mm by 24-mm samples (in duplicate) to approximately 10⁴ CFU/mm² with 70-50-11-0.5-Hex, 70-50-11-0.5, the QAC control, or the glass control for 4 h and shaken at 60 rpm at 30°C. For experiments longer than 4 h, moistened filter paper was placed into an empty well, separate from the test surfaces. The bacteria were stained according to the directions of the Live/Dead bacterial viability kit (BacLight; Thermo Fisher Scientific, Germany), directly in the medium, and imaged using a Leica Microsystems broadband confocal TCS SP5 inverted microscope with a 10×/0.4-numerical-aperture (NA) IMM objective (Leica, Germany). Images were analyzed in Fiji (79) in two ways: (i) by selective particle counts of the live (green) and dead (red) bacterial cells with the Analyze Particles protocol and (ii) by percent surface coverage per field of view by means of the averaged fluorescence intensity (InDEN).

JIS Z 2801/ISO 22916 contact thin-film assay.

In addition to the above-described assays, the antibacterial activity and the effect on the culturability of bacteria on the antiseptic surfaces were determined according to the JIS Z 2801 assay, also known as BS ISO 22196:2009, entitled Measurement of Antibacterial Activity on Plastics and Other Nonporous Surfaces (42). Cultures grown overnight were washed and resuspended in phosphate-buffered saline (PBS) to an OD of ∼0.01 with a spectrophotometer (Eppendorf BioPhotometer), to approximately 10⁵ CFU/mL. One hundred microliters of the adjusted culture was inoculated onto the samples and covered with 22- by 22-mm PE film (to maximize surface coverage and exposure). The experimental setup was housed in a nontreated 6-well plate (PS) (Costar, catalog no. 3736; Corning), with an empty well lined with moistened filter paper to maintain humidity overnight. Samples were incubated at room temperature for 20 to 24 h and washed with 5 mL of PBS at 220 rpm for 5 min. Viability counts were performed by serial dilution in PBS and plating onto LB agar plates (Sigma-Aldrich), in triplicate, and incubation at 30°C for 24 h. CFU were counted manually when possible and verified using the cell counter plug-in in Fiji (ImageJ) (79). The antimicrobial efficacy was recorded as the mean log (log_mean) reduction (LR) of the viable cell count, as follows: LR = (substrate log_final − control log_final)/(substrate log_initial − control log_initial).

Spacer experiment thin-film assay.

This modified live/dead staining assay assesses the effect of a reduced volume-to-surface-area ratio on the antimicrobial activity of surfaces. In addition, it can differentiate between surface-immobilized biocide killing and solution diffusion in either bioactivated or continuous-release systems. The original assay developed by Green et al. (38) utilizes an inoculation “vessel,” with the active surface and a coverslip (either blank or treated with a biofilm), separated by PS beads. In our case, these two surfaces were separated by an imaging spacer (Grace Bio-Labs Secure Seal, 120 μm, 9-mm diameter), allowing 3 distinct zones that can be analyzed by fluorescence microscopy: the substrate surface, the solution, and the top untreated surface. The test bacteria were diluted in modified M9* medium (we supplemented M9* minimal salts to prepare modified M9* medium with glucose and casein hydrolysate for final concentrations of 0.004% glucose and 0.02% casein hydrolysate) to higher concentrations of 10⁶ CFU/mL to 10⁷ CFU/mL (lower concentrations were not always detectable by microscopy on the test surfaces), and a Live/Dead bacterial viability kit was directly added to the test solution (BacLight; Thermo Fisher Scientific, Germany). Imaging spacers were securely adhered to the test substrate, into which 8.5 μL of a bacterial solution was pipetted; the adhesive foil was removed; and the setup was sandwiched with a cleaned (70% ethanol and dried) 24- by 24-mm glass coverslip. Images were taken every 5 to 8 min for 3 to 4 h in all 3 preset zones using a Leica Microsystems broadband confocal TCS SP5 inverted microscope with an enclosing custom-made incubation chamber (set to 30°C) and equipped with a 10×/0.4-NA IMM objective (Leica, Germany). The bacterial coverage was quantified by counting the live (green) and dead (red) bacteria as well as determining the percent surface coverage per field of view as described above using ImageJ.

Direct-contact, single-cell growth kinetics assay.

The direct-contact, single-cell growth kinetic assay allows direct contact between the surface-immobilized biocide and bacteria. A single colony of E. coli W3110 was inoculated overnight in 2 mL of M9* medium (modified and supplemented M9 medium [M9 minimal salts {catalog no. M6030; Sigma-Aldrich, Germany} supplemented with 2% casein hydrolysate broth {Sigma-Aldrich, Germany} and 0.4% glucose {20% glucose; Teknova, VWR Germany}]). The culture grown overnight was diluted to 1:200 in fresh M9* medium and grown to mid-exponential phase (10⁸ CFU/mL; OD₆₀₀ of ∼0.3). The “direct-contact” growth kinetic assay comprises a functionalized polymer (substrate) in direct contact with an M9* medium agarose pad (3%). A 20-mm by 20-mm by 3-mm agarose pad was firmly placed onto a glass 24- by 24-mm coverslip, after which a 3-μL volume of bacterial cells in the exponential growth phase was placed onto the agarose pad (OD of ∼0.3), incubated for 30 s, and then sealed with a cover glass slide with the test surface. The growth behavior of the immobilized bacteria was then observed microscopically at regular 5-min intervals for up to 4 h at 30°C, using a Zeiss (Jena, Germany) Axiovert 200 inverted microscope with an enclosing custom-made incubation chamber (set to 30°C) and equipped with an AxioCam 503 mono charge-coupled-device (CCD) camera (Zeiss, Jena, Germany) and a Zeiss EC Plan-Neofluar 40×/0.75-NA objective. The growth kinetics were quantified by monitoring the relative bacterial cell length for individual cells until the first cell division occurred using the MicrobeJ (80) plug-in of ImageJ (79).

Statistics.

Where possible, each experiment was performed in triplicate with at least 2 samples per experiment. The number of viable bacterial cells (viable counts) were quantified by Fiji (80), determined by comparing mean values and standard deviations, and a t test was performed in GraphPad Prism (GraphPad Software, San Diego, CA, USA). For log reductions, +1 was added to any zero values to allow assessment. When plotted as box-and-whisker plots, suspected outliers (>10% deviation) are plotted outside the outer quartile range.

Data availability.

All data are available from the corresponding authors upon request.