Mechanical Properties and Concentrations of Poly(ethylene glycol) in Hydrogels and Brushes Direct the Surface Transport of Staphylococcus aureus

Abstract

Surface-associated transport of flowing bacteria, including cell rolling, is a mechanism for otherwise immobile bacteria to migrate on surfaces and could be associated with biofilm formation or the spread of infection. This work demonstrates how the moduli and/or local polymer concentration play critical roles in sustaining contact, dynamic adhesion, and transport of bacterial cells along a hydrogel or hydrated brush surface. In particular, stiffer more concentrated hydrogels and brushes maintained the greatest dynamic contact, still allowing cells to travel along the surface in flow. This study addressed how the mechanical properties, molecular architectures, and thicknesses of minimally adhesive poly(ethylene glycol) (PEG)-based coatings influence the flow-driven surface motion of Staphylococcus aureus MS2 cells. Three protein-repellant PEG-dimethylacrylate hydrogel films (~100 μm thick) and two protein-repellant PEG brushes (8–16 nm thick) were sufficiently fouling-resistant to prevent the accumulation of flowing bacteria. However, the rolling or hopping-like motions of gently flowing S. aureus cells along the surfaces were specific to the particular hydrogel or brush, distinguishing these coatings in terms of their mechanical properties (with moduli from 2 to 1300 kPa) or local PEG concentrations (in the range 10–50% PEG). On the stiffer hydrogel coatings having higher PEG concentrations, S. aureus exhibited long runs of surface rolling, 20–50 μm in length, an increased tendency of cells to repeatedly return to some surfaces after rolling and escaping, and relatively long integrated contact times. By contrast, on the softer more dilute hydrogels, bacteria tended to encounter the surface for brief periods before escaping without return. The dynamic adhesion and motion signatures of the cells on the two brushes were bracketed by those on the soft and stiff hydrogels, demonstrating that PEG coating thickness was not important in these studies where the vertically oriented surfaces minimized the impact of gravitational forces. Control studies with similarly sized poly(ethylene oxide)-coated rigid spherical microparticles, that also did not arrest on the PEG coatings, established that the bacterial skipping and rolling signatures were specific to the S. aureus cells and not simply diffusive. Dynamic adhesion of the S. aureus cells on the PEG hydrogel surfaces correlated well with quiescent 24 h adhesion studies in the literature, despite the orientation of the flow studies that eliminated the influence of gravity on bacteria-coating normal forces.

Graphical Abstract

INTRODUCTION

Bacterial infections are a growing healthcare concern because of the increased use of polymer-based medical devices and growing bacterial resistance to antibiotics. Both temporary and permanent devices including catheters and orthopedic implants, respectively, are common infection sites. Conservative estimates attribute over 500 000 infections per year to the placement of catheters,¹ incurring an average financial burden of ~$30 000 per infection.² Growing incidences of methicillin-resistant Staphylococcus aureus (MSRA) infections³ have directed scientific interested toward S. aureus⁴ and driven research to develop strategies to prevent infection. Until recently, studies of bacterial adhesion and its control through engineered materials have focused exclusively on interfacial chemistry. This paper addresses the knowledge gap surrounding how the mechanical properties of a coating affect bacterial adhesion and travel along a surface in flow. We address the dynamic adhesion events associated with early biofilm formation, focusing on minimally adhesive coatings where weak reversible bonds permit cell motion and enable potential coupling with coating mechanics.

The possibility of bacterial sensitivity to the mechanical properties of a surface is new concept, only now starting to be explored for a few systems.^5–8 A positive correlation between coating modulus and microbial attachment after settling (and rinsing of loosely adhered microbes) has been reported for polyelectrolyte multilayers,⁹ agarose gels, polyacrylamide hydrogels, and poly(ethylene glycol) (PEG) hydrogels⁷ in isolated labs including one of our own. Others report an inverse correlation between substrate modulus and bacterial retention after settling for ultra-soft injectable hydrogels¹⁰ and poly(dimethylsiloxane).¹¹ Complicating the variety of perspectives is the likelihood that mechanics play out differently in different types of studies. For instance in flow, cell–surface contact times are typically short while in quiescent settling experiments, gravity compresses cells onto test surfaces and incubation times are ranging from hours to days. Furthermore, both physical and biological mechanisms may be involved. For instance, weak adhesion between cells and minimally adhesive coatings may be more evident in quiescent rather than flow studies.^12,13 Similarly, cell adhesion involving bonds with slow forward rate constants may be better detected in settling studies.

Along with the importance of longer contact time in quiescent versus flow conditions, the cells may respond to specific mechanical cues from a coating. For instance rate-dependent catch–stick bonds involving Fim-H increase Escherichia coli adhesion to mannose-presenting surfaces with increased shear.^14,15 Shear force also enhances the binding of E. coli to red blood cells through interactions with antigen I,¹⁶ whereas shear stress increases the adhesive residence time of Pseudomonas aeruginosa adhesion on glass.¹⁷ An influence of shear is also evident in the shear-enhanced adhesion of Staphylococcus epidermidis cell clusters to fibronectin,¹⁸ and the shear-dependent adhesion of S. aureus to collagen- and protein-coated surfaces.¹⁹

The cases above demonstrate that mechanical interactions couple with specific microbial adhesive interactions. Because rapid irreversible adhesion will mask the underlying influence of substrate mechanics, the current study focuses on surfaces that bind cells weakly. In particular, we consider how reversible cell interactions with “fouling-resistant” surfaces couple with substrate mechanics to direct surface transport of cells.

PEG coatings are the focus of this research because they are industrially relevant and their fundamental properties, biocompatibility, and sterically repulsive interactions are well-characterized. This study compares “nonbioadhesive” hydrated PEG coatings in the form of hydrogels²⁰ (crosslinked swollen films on the order 100 μm thick) and “brushes” (lawns of end-tethered solvated chains, order 10 nm thick),^21,22 shown schematically in Figure 1. PEG hydrogels must be properly designed by selecting a crosslinker, crosslink density, and extent of swelling that avoids protein and cell capture^23,24 Likewise in order to avoid protein and cell adhesion driven by the substrate or anchoring groups, PEG brush thicknesses and densities must be appropriately engineered.²⁵ The literature contains many reports that these goals can be successfully achieved.^22,26,27 The literature also describes cases where PEG layers support protein and cell adhesion because layers are too thin^25,28 or flawed,^29,30 or because anchoring and functional groups are exposed.²¹ The current work employs PEG-coating formulations and architectures described in the literature,^21,22,24 and confirmed here, to prevent protein adsorption and, in gentle flow, to avoid cell accumulation within detectible limits.

Figure 1.

Schematic of PEG hydrogel and PEG brushes on glass substrates. The mesh size of hydrogels and the correlation length of brushes are denoted by ξ.

While the PEG systems in the current work include optimally designed hydrogel coatings and brushes with no cell accumulation, weak adhesive interactions can originate from the PEG chains themselves. Well-hydrated PEG is a hydrogen bond acceptor and may bind weakly and reversibly with hydrogen bond donors.³¹ This is known to occur in water, where both the PEG and a hydrogen bond donor release previously hydrogen-bound water as they form hydrogen bonds with each other. Silica and, at low pH, organic acids³² are well-known examples. Teichoic acid on the surface of S. aureus may act similarly. Separately, PEG’s ether oxygens are electronegative, binding metal cations^33,34 and lysozyme³⁵ in solution. Additionally, otherwise protein-resistant PEG brushes bind lysozyme³⁰ and streptavidin.³⁶ These interactions are generally too weak to arrest bacteria on hydrated PEG surfaces in the gentle flow used in these studies, but along with the hydrophobicity of the carbon pairs of each PEG monomer unit, they may contribute to the ability of the body to produce antibodies against PEG.^37,38 Thus, PEG interactions, though weak and reversible, are biologically relevant and may couple with mechanics during bacterial interactions, the focus of the current work.

With new understanding now emerging about the role of mechanics in bacterial interactions with biomaterials, we demonstrate here that surface mechanics and coating water content control dynamic adhesion of flowing bacterial cells. We demonstrate how weak interactions between gently flowing S. aureus MS2 cells and biorepellant PEG hydrogel or brush coatings produce protracted coating-dependent dynamic cell–coating contact, including passes of rolling-type motion. The study examined bacterial interactions with three PEG hydrogel films of varied modulus (2–1300 kPa) and two PEG brushes with tether molecular weights 2000 and 5000 Da, whose moduli were in the same range as the hydrogels. The extent to which S. aureus cells rolled or escaped the surfaces was dependent on the coatings themselves and bacterial interactions were more protracted than were the 1 μm spherical microparticle controls. Specifically, S. aureus bacteria exhibited longer dynamic surface residence times and integrated periods of surface contact with the stiff hydrogels and with the higher molecular weight brushes than with the soft hydrogels.

MATERIALS AND METHODS

Model Bacteria Strain.

S. aureus MW2 (S. aureus) was a gift from Prof. Neil Forbes at the University of Massachusetts Amherst. This is a clinically relevant strain of community-acquired methicillin resistant S. aureus. S. aureus was grown overnight at 37 °C and harvested after a 16 h growth in tryptic soy broth (Sigma-Aldrich, St. Louis, MO). To remove residual proteins and other macromolecular constituents, we washed cells three times with phosphate buffered saline (0.008 M Na₂HPO₄, 0.002 M KH₂PO₄, and 0.15 M NaCl) before resuspending them in phosphate buffered saline. For all bacteria flow experiments, the resuspended cells (10⁸ cells/mL) were used within 1 h of preparation. Viability screening before and after flow experiments confirmed that the bacteria maintained viability throughout all experimental procedures. In the viability screening procedure, cells were stained in the dark with propidium iodide (Sigma-Aldrich, excitation/emission at 535 nm/617 nm) for 10 min to identify dead cells. A minimal fraction of dead cells was observed using a Zeiss Microscope Axio Imager A2M (Thornwood, NY).

Engineered Poly(ethylene oxide)-Coated Microparticles.

Control studies employed monodisperse 1 μm spherical silica microparticles from Gel Tech (Orlando, FL), with ~0.3 mg/m² of adsorbed 10 kDa poly(ethylene oxide) (PEO, Laysan Bio Inc., Arab, AL), prepared by mixing the microparticles and the polymer solution in appropriate amounts. (The saturation coverage of PEO on silica is0.35–0.4 mg/m²).³⁹ These PEO-modified microparticles exhibited a zeta potential of −15 mV, similar to that reported for S. aureus.^40,41 Dynamic light scattering revealed PEO-modified particle diameters that were indistinguishable from those of bare microparticles, indicating that the PEO adsorbed in a flat nanometer-scale layer dominated by trains (segments in contact with the surface) with few little tails or loops. In this way, the PEO, with its adsorption by hydrogen bonding to the nondissociated surface silanols on the silica microspheres,^39,42 blocked the majority of the hydrogen bonding sites on the microparticles that would otherwise bind to the PEG coatings within the flow chamber. The adhesion of uncoated microparticles to PEG brushes was the focus of a prior study.⁴²

Fabrication and Characterization of PEG Hydrogel Coatings.

PEG hydrogels were prepared according to previously established protocols.^7,43 Soft-1 and stiff-1 PEG hydrogels were formulated using solutions of 11 wt % and 55 wt % PEG dimethacrylate (M_n 750 g/mol, Sigma-Aldrich, St. Louis, MO), respectively, which were dissolved in phosphate buffered saline that was sterile-filtered. Soft-2 PEG hydrogels were prepared using a mixture of 5 wt % PEG dimethacrylate (M_n 750 g/mol, Sigma-Aldrich, St. Louis, MO) and 10 wt % PEG dimethacrylate (M_n 20 000 g/mol Polysciences, Warrington, PA) in phosphate buffered saline that was sterile-filtered. The radical photo initiator, Irgacure 2959 (0.8 wt %, BASF Ludwigshafen, Germany), was added to the polymer precursor solution to facilitate photopolymerization. The hydrogel precursor solution was then pipetted onto a glass coverslip (Fisher Scientific, Fair Lawn, NJ) that was functionalized with 3-(trimethoxysilyl)propyl methacrylate (Sigma-Aldrich) to covalently attach the hydrogel to the coverslip. A 24 mm × 40 mm glass coverslip (Fisher Scientific) was carefully placed on top of the hydrogel precursor solution to inhibit oxygen diffusion. Capillary forces held the precursor solution within the defined dimensions of the coverslip, which created hydrogels with a uniform thickness. Following polymerization, the top coverslip was removed with forceps and the glass-bound hydrogels were equilibrated for 48 h in 25 mL of phosphate buffered saline. Prior to experiments in the flow chamber, the hydrogel’s length and width were trimmed using a sterile razor to fit the rubber gasket that sealed the chamber.

The thicknesses of PEG hydrogel coatings were determined using a digital micrometer (Mitutoyo Corporation, Kawaski, Japan) by averaging five measurements on each of at least three fully swollen hydrogels.

Equilibrium swelling experiments were performed by using free-standing hydrogel specimens to determine the hydrogel’s volumetric swelling ratio, Q, and equilibrium PEG concentration. Hydrogels were swollen in phosphate buffered saline for 48 h at 23 °C until thermodynamic equilibrium swelling was achieved, then weighed to obtain their equilibrium swelling mass, M_S. The specimens were then lyophilized (Labconco, FreeZone Plus 2.5 Liter Cascade Console Freeze-Dry System, Kansas City, MO) for 72 h and weighed to determine their dry mass, M_D.

The equilibrated polymer concentration is the inverse of Q, which was defined as M_S/M_D. A modified Flory theory⁴⁴ was then applied to determine the mesh size, ξ, from eq 1

$$\xi =v_{2,s}{}^{-1/3}{\left({\overline{r}}^2\right)}^{1/2}$$ (1)

where υ_2,s is the swollen volume fraction of the polymer and ${\left({\overline{r}}^2\right)}^{1/2}$ is the average end-to-end distance of the crosslinked PEG. Four replicates were tested for each hydrogel. Table 1 summarizes the polymer concentration and swelling ratio of PEG hydrogels.

Table 1.

Fabrication of PEGDMA Hydrogels

	PEG content (wt %)
name	Mn: 750	Mn: 20 000	swelling ratio
soft-2	5	10	15 ± 0.40
soft-1	11	NA	8.6 ± 1.0
stiff-1	55	NA	2.2 ± 0.05

To determine the hydrogel moduli, small amplitude oscillatory shear (SAOS) experiments were conducted using a plate–plate geometry, with a diameter of 20 mm and a gap of 1 mm (Kinexus Pro Rheometer, Malvern Instruments, UK). Freestanding hydrogel specimens (circular, 25 mm diameter × 1 mm height) were loaded into the rheometer and trimmed to size using a razor blade. A strain amplitude sweep was performed to ensure that experiments were conducted within the linear viscoelastic region and a strain percent of0.1% was selected. Oscillation frequency sweeps were conducted over an angular frequency domain, 1.0 and 100 rad/s at 23 °C.

Fabrication and Characterization of PEG Brushes.

Poly-l-lysine hydrobromide (PLL)–PEG copolymers were synthesized, as described previously.^21,22,45 PLL with a nominal molecular weight of 20 000 g/mol (Sigma-Aldrich, St. Louis) dissolved in 50 mM sodium borate buffer (pH 9.1), was reacted with either a nominal 2k molecular weight n-hydroxysuccinimidyl ester of methoxypoly-(ethylene glycol) acetic acid or a nominal 5k molecular weight PEG sodium valeric acid (Laysan Bio Inc., Arab, AL). The relative amounts of PLL and PEG were adjusted so that the PLL was approximately one-third functionalized. Following the reaction, the copolymers were purified by dialysis against phosphate buffered saline and then in deionized (DI) water. After freeze drying, the samples were stored at −20 °C. Copolymers in D₂O were characterized using ¹H NMR on a Bruker 400 MHz instrument. The extent of PLL functionalization was determined from the relative areas of the lysine side chain peak (−CH₂−N−) at 2.909 ppm and the PEG peak (−CH₂−CH₂−) at 3.615 ppm.

Substrates were prepared by immersing FisherFinest microscope slides overnight in concentrated sulfuric acid and rinsing thoroughly in DI water to produce a clean silica surface. Then to deposit a polymer brush, the slides were immediately sealed in a laminar flow chamber and contacted with flowing phosphate buffer. Next, a 100 ppm copolymer solution in 0.01 M phosphate buffer (0.008 M Na₂HPO₄ and 0.002 M KH₂PO₄) was flowed over the surface for 20 min at a wall shear rate of 5.0 s⁻¹. The flowing phosphate buffer was subsequently replenished for ~10 min. The copolymer adsorption traces were frequently monitored by near-Brewster reflectivity³⁹ to confirm the copolymer adsorption, near 1.1 mg/m² for either copolymer. Knowing the total adsorbed amounts and PEG contents in the layers allowed us to describe the brushes, in the Results section, in terms of parameters from established models for brushes. The calculations are reviewed in the Supporting Information.

Test of Protein Repellance.

To confirm that brushes and hydrogel coatings were properly prepared with respect to protein repellance, fibrinogen was employed as a probe for the brushes and fluorescently tagged FITC-fibrinogen, prepared in-house as previously described,⁴⁶ was employed as a probe for the hydrogels. In the case of the brushes, 100 ppm fibrinogen in phosphate buffer was flowed over brush surfaces in the shear flow chamber, and protein adsorption was confirmed to be zero, within the 0.01 mg/m² resolution of the near-Brewster reflectometer.²⁶ FITC-fibrinogen was similarly flowed over the hydrogel layers and scattering from the gel registered a fluorescent signal from the fluorescent protein in the bulk solution; however, upon reinjection of flowing buffer, the signal dropped back to the baseline, indicating no fibrinogen retention in or on the gels.

Interactions of Flowing Bacteria.

Bacteria flow studies were performed in a laminar flow chamber (240 × 178 μm), mounted in a custom-built lateral microscope that oriented the surface of interest perpendicular to the floor. This eliminated gravitational forces between flowing cells or microparticles and the wall. A syringe pump was employed to flow each suspension of ~10⁸ cells or microparticles/milliliter at a wall shear rate of ~15 s⁻¹. Video footage was recorded using a 20× Nikon objective and microparticle or cell motion was analyzed using the manual particle tracking feature in FIJI Image J v1.48 software (National Institutes of Health, Bethesda MD). A time step of 0.2 s was employed during particle tracking to calculate average velocities from the distances traveled by cells or particles during each time step.

At least 20 and up to 50 bacteria or microparticles, all those that qualified in a fixed time period, were tracked in each run. A “run”, in which a bacterial suspension was flowed through the chamber, typically lasted ~10 min. The numbers of cells or microparticles analyzed was somewhat variable because the run time itself was fixed. All qualifying microparticles or cells within a set run time, for that microparticle–coating combination period were analyzed. This ensured that there was no bias in the choice of microparticles. For instance, a cell or microparticle was tracked if it met the criteria for surface engagement, traveling less than 5 μm/s for a distance of 5 μm or greater, during a 10 min window of a run. (This criterion is motivated by the treatment of Goldman et al. for a sphere near in a wall in shear flow,⁴⁷ and corresponds to microparticles rotating ~1.5 times during surface engagement.) Notably, while large numbers of cells were pumped through the chamber only a very small fraction was near the coating and only a fraction of these collided and engaged the coating in a way that demonstrated dynamic adhesion as defined by the criterion. We found no significant differences between multiple runs with a given microparticle–coating choice. Data for 2–3 runs for each coating/microparticle-or-cell pair were combined in the result plots to produce the most meaningful statistics in comparing coatings and bacteria or microparticles. Additionally, control experiments employing microparticle suspensions 100 times less concentrated than regular conditions produced statistically identical values for the quantities measured.

RESULTS

Properties of PEG Hydrogel Films and Brushes.

Table 2 summarizes the hydrogel and brush properties. While similar in chemistry and having overlapping ranges of moduli and PEG concentrations, the PEG brush and hydrogel coatings differ in thickness by many orders of magnitude.

Table 2.

Characteristics of PEG Hydrogels and Brush Coatings

Sample name		equilibrated PEG content (wt %)	G′ or G (brushes) (kPa)	mesh size or persistance length (nm)	thickness (μm)	fibrinogen adsorption (mg/m2)
hydrogels	soft-2a	7 ± 0.1	1.8	4.1 ± 0.4	120 ± 5	<0.01
	soft-1	8.6 ± 2	9.5	2.7 ± 0.1	110 ± 5	<0.01
	stiff-1	46 ± 1	1300	1.0 ± 0.1	100 ± 5	<0.01
brushes	2k brush	11 ± 2b	1800b	1.9 ± 0.1c	0.007–0.009b	<0.01
	5k brush	6 ± lb	450b	2.9 ± 0.1c	0.015–0.017b	<0.01

^aBimodal network chain distribution, see Table 1.

^bCalculated as described in the Supporting Information, and following Gon, Fang, Santore, Macromolecules 2011.³⁰

^cTaken as the average distance between tethering points, based on the measured amount of copolymer adsorption and known PEG molecular weight.

The storage moduli of the PEG hydrogel coatings vary from 2 to 1300 kPa, whereas the compression moduli of the brushes vary from 450 to 1800 kPa, in Table 2. The hydrogels are identical to samples studied by Kolewe et al.⁷ but characterized here via SAOS rheology. Indeed, dynamic oscillatory rheology is a standard approach for the characterization of hydrogel materials and provides quite similar values to those previously determined by nanoindentation experiments using hydrated atomic force microscopy. Estimates for brush thicknesses and moduli are calculated based on the Alexander–de Gennes treatment of the Flory brush,^29,48–50 detailed in the Supporting Information. This frequently employed approach provides excellent first order property estimates from known quantities such as the overall mass and second virial coefficient. The ranges of moduli for the hydrogel and brush coating are overlapping, allowing further consideration of coating thickness and architecture itself. Hydrogel coatings are on the order of 100 μm thick while the brushes are the molecular scale, on the order of 10 nm. The hydrogels also contain crosslinked acrylate moieties every 750 or 2000 Da of molecular weight, whereas the brushes are comprised of pure PEG tethers with OH end groups. Notably, the additional functionality within the hydrogel is sufficiently shielded by the PEG chains to avoid fibrinogen adsorption.

Table 2 includes estimates for the average PEG concentration in the hydrogels and brushes, with the stiffer coatings having a higher PEG content. The range of average PEG concentration in the hydrogels overlaps that in the brushes. Also similar between the PEG hydrogels and brushes is the range of mesh sizes. For hydrogels this is the average calculated distance through space between crosslinks. For brushes the analogous quantity is the correlation length, equal to the average distance between chain-anchoring points, following directly from the measurements of the adsorbed amount of PEG in the brush and the known tether molecular weight.²⁹ Clear in Table 2, mechanical and physical properties of the coatings fall within overlapping ranges, even though the architectures and thickness of the hydrogels and brushes are vastly different.

A key property for all five of the coatings is their lack of fibrinogen adsorption within detection limits of 0.01 mg/m², via fluorescence and reflectometry, confirmed in this study and consistent with literature.^7,26,51 As fibrinogen is among the more adhesive of serum proteins^21,22 (more than albumin⁴⁶), it is commonly employed to screen for potential serum interactions.⁵² Lack of arrest and retention of flowing S. aureus bacteria was an additional criterion for coatings in this study, allowing us to examine the impact of weak reversible bacteria–PEG interactions. Worth noting, the lack of fibrinogen adhesion argues against the possibility where crosslinking and other sources of heterogeneity lead to adhesive surface features. Also, because the gels are prepared very near to their final swollen composition, the swelling ratio in the experimental section represents the water fraction and not a start-to-finish change of the gel. As a result, one would not expect buckling and wrinkling and we saw no evidence of mechanical surface instabilities by microscopy.

Surface-Associated Motion Signatures.

Of the millions of cells or microspheres that flowed through the millimeter-thick chamber during a run, only those near the test surface were within focus of the microscope. A fraction of these in-focus cells or microparticles encountered the coated wall incidentally through diffusion and, without strong adhesion, diffused back into faster streamlines further from the surface. Though cells or microparticles did not arrest or accumulate appreciably on the coatings, their weakly attractive interactions with the coated wall were evident in the form of transient dynamic adhesive “engagements”. That is, reversible adhesion caused cells or microparticles to travel nearest to the surface for longer periods and at slower velocities than free microparticles.

Figure 2A provides example trajectories of two S. aureus cells flowing over a 2k brush. Cell 1 took 18 s to traverse the 240 μm field of view, whereas cell 2 took 37 s. Both had periods of fast and slow movement, with the slow portions nearly similar in slope, giving velocities in the range 3–5 μm/s. For cell 2, Figure 2B shows the velocity at each point. The traces for cell 2 contain two sections of low, nearly constant velocity, in which the cell underwent a surface-associated motion, which we term “engagement”. During engagement, cells or microparticles progress or roll along the surface in flow more slowly than nonengaged near-surface cells or microparticles. Free cells and microparticles could additionally traverse faster streamlines, further from the surface. Thus, the cells or microparticles move toward and away from the surface except when they are adhesively engaged. Of note, in Figure 2A, the slower moving portions of the faster cell also qualify as engaged, with a similar velocity in the range 3–5 μm/s. Because rotation was not able to be consistently quantified due, in part, to the small cell/microparticle size, we cautiously refer to the dynamic adhesion as “engagement” rather than “rolling.” However, the vorticity of the shear field, with faster streamlines away from the wall, generally imparts cell or microparticle rotation.

Figure 2.

(A) Example trajectories of two representative S. aureus cells flowing near a 2k brush. All cells travel horizontally across the chamber and their distance traveled increases on the y-axis of this plot as a function of time on the x-axis. Regions of more gradual rise in this graph indicate slower cell progression across the surface. (B) Instantaneous velocities, for S. aureus cell 2 from part (A). Two periods of engagement with a rolling-like behavior are indicated.

Known features of cell rolling provide further guidance to interpret cell and microparticle behavior. The velocity of a free sphere in shear flow at a specified distance from the wall was quantified by Goldman et al.⁴⁷ For a 1 μm sphere in flow having a wall shear rate of 15 s⁻¹, its velocity will be 4–5 μm/s when the sphere is separated from the wall by 1–2 nm (approaching the correlation length of the brushes and hydrogels). Per Figure 2, we considered that microparticles or cells moving more slowly to be adhesively engaged with the PEG coatings through reversible interactions such as hydrogen bonding. Rolling motion enables simultaneous bond formation at the front of the cell or microparticle and disbonding in the rear while the cell or microparticle continues to move along the wall. The flow vorticity produces rotation, though rolling motion may be unsteady (resembling skipping at times shorter than the video framing rate) because the rotation itself is not visible for 1 μm spheres. Escape or disengagement requires breaking all contact, and so is stochastic. Some reversibly adhered microparticles or cells will travel farther and have longer interfacial residence than others. Therefore in addition to the 5 μm/s criterion to distinguish surface engagement, we imposed a practical criterion in which each microparticle or cell appear surface-engaged for a full second in order to be counted as engaged in subsequent analyses. This practical choice involves 1.5 cell rotations in contact with the surface. Thus, in this work we quantify the relatively long-lived cell–surface dynamic interactions, those with the chance of producing a biological response compared with interactions of shorter duration.

Distribution of Lengths for Individual Engagements.

For each run in this study, we identified all the qualifying surface engagements of cells or microparticles. The surface engagements for S. aureus cells in Figure 2 are representative examples. For each engagement, we determined its length and duration in time. Figure 3A summarizes the length distributions of the engagements for S. aureus cells with each surface type. Part B contains analogous data for the PEG-coated silica microparticles. However, some cells or micro-particles sometimes exhibited multiple engagements in a single trace, the distribution of engagement lengths, reported in Figure 3 was no different between the first and subsequent encounters in the observation window. Additionally, while multiple surfaces of each type were employed to obtain large numbers of engagements, we found no significant differences in the data acquired on multiple surfaces of the same type.

Figure 3.

Engagement length distributions of (A) S. aureus and (B) PEG-coated silica microparticles on each PEG coating. The frequency of occurrence on the y-axis represents the percentage of the total number of surface encounters identified. Typically, 25 or more cells or microparticles were tracked in each run, and 2–3 surfaces of each type were studied with microparticles or cells. This gives at least 60 cells or microparticles on each type of coating.

Figure 3 reveals material dependence for the distributions of the S. aureus engagement lengths on PEG-based coatings, despite their common chemistries. S. aureus cells rolled or exhibited longer lengths of dynamic adhesion on the stiff hydrogel and on the PEG brushes compared with either of the two soft hydrogels. In fact, on the soft hydrogels 90% of the engagements were shorter than 25 μm while on the stiff hydrogels about one-third of the rolling engagements were longer than 25 μm. Also in Figure 3, the S. aureus cells were adhesively engaged for greater distances than were the PEG-coated microparticles. These monodisperse 1 μm silica microparticles, coated at 80% of saturation coverage with adsorbed PEO in order to block enough hydrogen bonding sites so that microparticles did not arrest/accumulate on the PEG coatings, serve as a benchmark for the dynamic behavior of similarly sized and shaped cells. The difference between cells and microparticles, both too weakly adhesive to arrest or accumulate on any of the coatings, demonstrates distinct adhesive differences in near-surface cell–particle motion. The bacterial cells, though nonarresting in weak flow, are clearly more adhesive than the microparticles. The differences in the behaviors of the cells and microparticles argue that the residence time distributions seen for the cells are not a natural consequence of near surface diffusion in flow because both the cells and microparticles are 1 μm spheres.

Repeat Surface Engagements.

Evident in the example of Figure 2B is that a dynamically engaged cell can leave the surface and later re-engage. For 10 min of video for three surfaces of each type, Figure 4 summarizes the numbers of engagements per cell or particle in the 240 μm-long field of view, for cells or particles having at least one engagement. In terms of the numbers of engagements per cell, S. aureus cells exhibited a remarkable ability to adhesively distinguish the different PEG coatings. Substantial differences between cells and engineered particles are also clear from Figure 4A versus 4B. Most striking, on the stiff hydrogel but not on the soft hydrogels, S. aureus cells exhibited a clear preference (with 99% statistical certainty, different from the other runs) to engage the surface more than once. On the 5k brush, S. aureus had only a slight preference for a single encounter (as opposed to a different specific number) within the field of view, but multiple engagements were prevalent, occurring for more than 60% of the encounters. Here a substantial fraction of the cells that engaged (15%) still adhesively engaged the surface 4 times in the field of view. The soft hydrogels and the 2k brushes, by comparison, produced more single encounters and the probability of multiple engagement decreased with the engagement number, with less than 10% of cell encounters producing four rolling-type engagements.

Figure 4.

Distribution of the frequency of multiple surface engagements for (A) S. aureus cells and (B) PEG-coated silica microparticles. These data summarize the runs from Figure 3. The x-axis refers to the total number of observed engagements of each cell/microparticle within the viewing area of length 240 μm. An asterisk (*) denotes statistical significance (p = 0.01) within an engagement count.

In Figure 4B the engineered microparticles interact, in terms of numbers of engagements, nearly independent of coating type, compared with the cell behavior in Figure 4A. Also, the particles exhibited a slight tendency for fewer overall engagements compared with the S. aureus cells: there were fewer instances of microparticles encountering the coating 4 times in the field of view, and these occurred on the hydrogels and not the brushes.

Cumulative Cell–Surface Contact.

Differences in the engagement lengths (Figure 3) and in the numbers of engagements per cell or particle in the field of view (Figure 4) translated to differences in the overall dynamic contact, on a per-cell or per-microparticle basis, as summarized in Figure 5. Here, for each cell or microparticle that engaged the surface, the total time of adhesive engagement in the field of view was determined, as defined as travel at a velocity ≤5 μm/s for at least 1 s. This metric includes multiple surface engagements per cell or microparticle. For reference, it takes approximately 45 s for a 1 μm sphere to traverse the field of view at 5 μm/s, motivating the scale on the y-axes. The analysis does not include cells or particles which rolled into or out of the field of view, eliminating a modest population having longer engagement times.

Figure 5.

Total residence time per individual (A) S. aureus cell and (B) PEG-coated microparticle on PEG coatings. The residence time for each cell or microparticle is the sum of all surface engagements, resulting in the total residence time in the observation window. The solid black line in each box represents the median residence time while the top and bottom of each box represent the 25th and 75th quartile distribution, respectively. (A) S. aureus cells experienced significantly (99% certainty) longer residence time on the stiff-1 hydrogels than on all other hydrogels and brush surfaces. There is also a significantly longer (99% certainty) residence time of S. aureus cells on the 5k brush than the soft-2 and soft-1 hydrogels, whereas the same residence time was observed on both soft hydrogels and the 2k brushes. There is a significant difference (95% certainty) between the 2k and 5k brushes. (B) Statistically the microparticles experienced a longer residence time on the hydrogels than on the brushes; though there is no difference between microparticles on the two hydrogels or between microparticles on the two brushes. One asterisk (*) denotes 95% significance and two asterisks (**) denote 99% significance between systems.

Figure 5A demonstrates substantial dynamic contact duration for the S. aureus on the PEG coatings, a striking behavior given that no cells or microparticles were captured or arrested during the run. The bacterial surface residence, albeit dynamic, is often statistically longer than that of the microparticles. Indeed the microparticle interaction with the brushes might be used as a lower adhesive benchmark: the short 1–2 s residence times indicate limited interaction at most, perhaps longer than purely diffusive dynamics, but still very weak. The microparticles were slightly more adhesive on the hydrogels than on the brushes. S. aureus cells had the longest integrated dynamic contact on the stiff PEG hydrogel, with cells engaging the surface for a median time of 16 s: these cells rolled approximately half the length of the flow chamber in no more than four engagements. On the stiff hydrogel, the top quartile of cells had contact in excess of 25 s across the 240 μm field. The second most adhesive surface was the 5k PEG brush, with the median cell rolling time of 12–13 s. The surfaces supporting the shortest cell residence were the soft hydrogels.

DISCUSSION

Adhesion Sensitivity to Different PEG Coatings.

This work demonstrated that while adhesive interactions may be too weak to arrest gently flowing S. aureus cells on PEG coatings, they can be sufficiently strong to alter the cell motion, increasing the near-surface cell residence and producing stretches of rolling-type motion that maintain dynamic bacteria–surface contact. Because S. aureus is nonmotile, this surface-associated cell motion in flow constitutes a mechanism of bacterial surface translocation that might lead to the spread of infection from implanted medical devices.^53–56 In contrast, stronger cell–coating interactions could arrest cells, localizing infection if the bacteria remain viable. Because of their large water content the hydrated PEG coatings comprise model systems that may interact differently with cells compared with classical solid implant materials such as titanium, polypropylene, or silicones.

Control studies with nonarresting-engineered microparticles, similar in size and shape to S. aureus cells, lacked the longer rolling engagements and, especially on the stiff hydrogels, the tendency to return to the surface that were seen for the bacteria. The differences between cells and microparticles established that the extended near-surface S. aureus dynamics resulted, at least in part, from attractive interactions or reversible bonds, that is hydrogen bonding (including exchange or release of previous bound waters) or weak electrostatic interactions involving the electronegative ether oxygen on the PEG, rather than hindered near surface diffusion that is seen for nonadhesive spheres near hard walls. Indeed, the interaction of the microparticles with the PEG brushes provided a lower adhesive limit in this study. Worth noting, the engineered microparticles contained 80% of the saturation amount of PEO chains,³⁹ leaving some hydrogen bonding sites available for the adhesion of additional PEG molecules (from the coating).⁴² It is possible that, with more PEO on the microparticles, they would have experienced even shorter residence time on the brushes. Notably, bare microparticles were irreversibly captured, in a separate study⁴² on PEG brushes identical to those studied here. Thus, loss of PEG chains from the particles during the current experiment is unlikely to have occurred because loss of PEG chains would have facilitated particle arrest on the PEG coatings.

The sensitivity of the S. aureus adhesion to differences in the PEG coatings is fascinating and unanticipated because all five coatings share chemical similarities, are protein repellant, and do not arrest S. aureus cells or microparticles. The stiff, more concentrated hydrogel supported the longest rolling stretches and substantial repeat engagements per cell, giving significantly more extensive overall dynamic contact compared with soft hydrogels. The 5k brush also supported extensive dynamic cell contact. This dynamic cell adhesion was more protracted than the interactions of these materials with the engineered microparticles. By contrast on the soft hydrogels, both the S. aureus cells and the microparticles had more nearly similar and less protracted dynamic interactions. While the effect of biomaterial stiffness on mammalian cells has been extensively studied and supported with biochemical mechanisms,^57,58 the understanding of the effect of biomaterial stiffness on bacterial cells is comparatively in its infancy. For instance, the biochemical mechanisms for some bacterial response to environmental stiffness are only recently emerging.^59–62 The ability of nonmotile bacteria to sense the mechanical properties of a surface without the use of extracellular organelles is a new proposition.

Potential Physical Interaction Mechanisms.

In this study, the S. aureus cells adhesively discriminated the different PEG coatings in dynamic conditions and at short times, motivating us to consider the physicochemical nature of cell–surface interactions on these timescales.

For a flowing suspended microsphere (a spherical cell or microparticle) to engage in continuous dynamic adhesion, for instance rolling, it must first reversibly adhere to a surface. This initial interaction involves adhesive groups across an interface at short times. Particle or wall deformation is not needed for this initial binding step and, indeed, previous studies provide examples of conditions where deformable particles are captured from flow at the same rates as rigid particles,⁶³ suggesting that any deformations, if they occur, might not influence capture. In the current study, it is unclear if deformation does or can occur on the timescales of particle–surface approach. We therefore expect that the influence of surface mechanics, may be more important in steps following the initial binding of a flowing particle or cell with the surface.

After a particle or cell adheres initially, rolling or other sustained dynamic surface engagement can be facilitated by the flowing fluid, which imparts a tangential force.⁶⁴ For a 1 μm sphere and a wall shear rate of 15 s⁻¹, this amounts to 0.12 pN.⁴⁷ As the cell or particle moves, new bonds are formed and existing bonds break. As long as the formation of new bonds in the front of the sphere is fast (and therefore not rate limiting), the rolling velocity will depend on the breaking of bonds in the rear of the sphere. This disbonding may dissipate energy by working the hydrogel or the bacterium, or by stretching PEG tethers of the brush. Additionally, deformation of soft materials can increase the contact area between a sphere and a surface. A 1 nm difference in the positioning of a 1 μm sphere against a brush, for instance, can increase the contact area by a factor of 2, in the Supporting Information. Therefore, even the mechanics of the brushes have the potential to influence cell rolling with minimum perturbation of coating surface. The substrate beneath a ~10 nm brush need not be mechanically involved, though it might be.

One expects that, compared with a stiff hydrogel, a soft hydrogel would more easily deform to increase contact with a microparticle or cell after the initial adhesion. Thus, rolling and slower interfacial travel would be facilitated by material deformations, which also reduce the particle or cell escape from the surface. (For reference, the soft hydrogels have lower moduli than S. aureus cells. The S. aureus MW2 cell moduli falls in the range 2–7 MPa.⁶⁵) The observed relatively short encounter lengths and more facile escape of cells from the soft hydrogels oppose this intuition. However, if one considers that the soft hydrogels are also lower in PEG concentration (Table 2), one might expect that fewer hydrogen bonds, for instance, are involved in individual cell interactions. This increases the probability that all bonds break at once, allowing earlier escape and producing shorter overall encounter lengths. This idea is consistent with the observed limited engagements of cells on brushes, which are relatively dilute.

The tendency for cells return to the stiffer coatings, especially the stiff hydrogel, is the most surprising of our findings. The velocities of cells during periods of escape suggest, based on our implementation of the model of Goldman et al.,⁴⁷ in spreadsheet calculations for our system parameters, that they are 100 nm or more from the surface, and they travel many microns before returning to the surface. Because single S. aureus cells are not known to form long adhesive tethers during their brief surface contact, we can only hypothesize the mechanism of the observed large travel distances prior to returning to the surface. It may be that the irregular structure of the hydrogel’s crosslinks creates the possibility of PEG tethers exceeding the molecular weight of the oligomers, containing multiple PEG-DMA oligomers, at the surface of the stiff hydrogel. (This cannot happen on the brush.) It may also be that the crosslinks in the hydrogels facilitate chemical interactions beyond those of the PEG and the cells or that combined adhesion and cell deformations produce hydrodynamic forces that favor cell return to the surface.

Despite bacterial sensitivity to subtle material differences, there is a great overarching similarity between cell interactions with hydrogels and brushes. The contact dynamics of the cells on the brushes tend to be bracketed by those on the hydrogels, with the stiff hydrogels being overall more adhesive and the soft hydrogels promoting less adhesive contact than the brushes. Thus, while bacterial adhesion distinguishes important material difference in these systems, it is clear that the bacteria experience the thin ~10 nm PEG brushes as a hydrated PEG material, similar to a hydrogel, rather than a rigid substrate. Despite the 5-order difference in the thickness of the PEG brushes and 100 μm hydrogel films, the similarities within the first few nanometers of the coating appear more important than overall coating thickness at the timescales of adhesive encounter and without gravity to enhance the cell–material contact.

Static Versus Dynamic Adhesion.

The dynamic adhesion of S. aureus cells on the hydrogels in the current study, employing flow and not having gravitational forces toward the hydrogel, correlates well with the retention of the same S. aureus strain in classical quiescent assays with the same PEG hydrogels. Kolewe et al. previously demonstrated that S. aureus cells, settled under gravity onto PEG hydrogels for either 2 or 24 h and then rinsed, were retained in greater numbers on the stiffer, more concentrated hydrogels.⁷ That study imposed different physics from the current paper, including a much longer quiescent cell–material contact time, independent of whether adhesion was occurring. By contrast, in the current study, an initial adhesion event must first occur, and only cells that achieve this threshold adhesion have the opportunity to roll. Additionally in Kolewe et al., gravitational forces pressed cells onto the hydrogel surface, potentially deforming both.⁸ By contrast, in the current study, gravitational forces acted tangential to the hydrogel surface. Thus, in Kolewe et al., there was more opportunity for molecular interdigitation to increase chemical bonding and for cell deformation to increase contact area. The same adhesion and cell deformation processes may be occurring in both studies, initiating and sensed at short times in our flow chamber, but persisting and influencing static adhesion in the quiescent protocol.

The current observations are also consistent with reports of the accumulation of individually reversible tethers during the adhesion of cocci bacteria to glass during settling.⁶⁶ When many of these bonds accrue, the bacterial cell becomes irreversibly bound. In the current case of flow, reversible tethers may produce engagement events but, because of cell motion, sufficient bonds to arrest cells may not form. Reversible adhesion events have, however, been implicated in increased colonization of E. coli.^67,68 Thus this flow study may represent a rapid screening assay or on-line sensors anticipating longer time behaviors. Additionally, the potential for biofilm formation may be evident even in these early behaviors where flowing cells associate dynamically with surfaces.

CONCLUSIONS

This study examined the dynamic adhesion of flowing S. aureus cells over a series of hydrated PEG coatings with varied modulus, PEG concentration, and correlation length. Two classes of coatings were compared: hydrogels on the order of 100 μm thick and end-tethered brushes of order 10 nm thick. Both the brushes and hydrogels overlapped in their range of mechanical properties and PEG concentrations and, importantly they were fully resistant to the adhesion of proteins or the arrest of flowing S. aureus cells.

Though flowing S. aureus cells did not arrest on any of the PEG coatings, they did exhibit dynamic adhesion in the form of protracted runs of near-surface motion best explained by rolling and intermittent contact, with runs of this dynamic extending for lengths in excess of 25 μm for 1/3 of the cells on the stiff hydrogel and brush surfaces, with shorter engagements observed on the soft hydrogels. On the stiff hydrogel and one of the brush surfaces, the escaped cells returned to re-engage the surface multiple times. Thus, the overall time spent in contact with a 240 μm run of surface in the field of view was the greatest on the stiff hydrogel, an average of 20 s.

The less extensive contact of control particles with the same PEG coatings suggests that physical–chemical interactions between the bacteria and the coating, rather than simply near surface diffusion in flow, were responsible for the extensive dynamic bacteria–coating contact. Therefore in addition to the role of mechanics, the greater adhesion of the stiffer hydrogels may be a result of the higher PEG concentration and greater opportunity for chemical interaction with PEG chains. Additionally, the similarities between cells engaged with the hydrogels or the brushes demonstrated that coating thickness was unimportant and that only the outermost regions of either coating were involved dynamic cell adhesion. Notably, in the current studies flow produces short contact times compared with classical settling protocols. Also gravitational contributions to bacteria–coating normal forces were absent in the current work, in contrast to settling-based protocols in which gravitational forces enhance interactions. Clearly in nature and technology surface orientation relative to gravity is varied.

The findings from this study, which focused on dynamic adhesion in which material points of the cell and the coating were in local contact for less than a second, ran parallel to separate studies indicating that the same cells were better retained on the stiff hydrogels compared with the soft hydrogels after several hours of settling and contact in a gravitational field. Thus, this work is a further demonstration that even in flow conditions that reduce bacteria–surface residence time, and without gravitational forces, weak interactions can influence near surface transport in ways potentially relevant to infection and biofilm formation. The observation of dynamic bacteria adhesion in flow may be a meaningful predictor of bioadhesion and biofilm formation in separate quiescent conditions. In particular, the ability of bacteria to mechanically discriminate surfaces without extracellular organelles is a new observation. The study may also have implications for flow drive bacterial travel along weakly adhesive coatings that aids the spread of infections.