Article pubs.acs.org/Langmuir
Diffusing Colloidal Probes of kT-Scale Biomaterial−Cell Interactions Gregg A. Duncan,† Sharon Gerecht,† D. Howard Fairbrother,‡ and Michael A. Bevan*,† †
Department of Chemical & Biomolecular Engineering, and ‡Department of Chemistry, Johns Hopkins University, Baltimore, Maryland 21218, United States ABSTRACT: In the optimization of applied biomaterials, measurements of their interactions with cell surfaces are important to understand their influence on specific and nonspecific cell surface adhesion, internalization pathways, and toxicity. In this study, a novel approach using dark field video microscopy with combined real-time particle and cell tracking allows the trajectories of biomaterial-coated colloids to be monitored in relation to their distance from cell perimeters. Dynamic and statistical mechanical analyses enable direct measurement of colloid−cell surface association lifetimes and interaction potentials mediated by biomaterials. Our analyses of colloidal transport showed polyethylene glycol (PEG) and bovine serum albumin (BSA) lead to net repulsive interactions with cell surfaces, while dextran and hyaluronic acid (HA) lead to reversible and irreversible association to the cell surface, respectively. Our results demonstrate how diffusing colloidal probes can be used for nonobtrusive, sensitive measurements of biomaterial−cell surface interactions important to therapeutics, diagnostics, and tissue engineering.
■
INTRODUCTION
The lower force limit of such mechanically based measurements is limited by the spring constant of the cantilever to ∼10 pN.19 It would be desirable to have a measurement method that is both direct and capable of resolving weaker biomacromolecular interactions on the order of the thermal energy, kT. To work toward this goal, diffusing colloidal probe microscopy (DCPM) has been developed as a method to nonintrusively measure interfacial interactions from trajectories of freely diffusing colloidal particles whose surfaces have been modified by adsorbed or grafted macromolecules near macromoleculecoated surfaces. With no external manipulation, measurements of potential energy versus position have been made with kTscale sensitivity and high spatial resolution. DCPM has been used in prior studies to characterize nonspecific and specific biomacromolecular interactions on model substrates.20−22 We have expanded this capability to simultaneously track particles and cells, which we implemented in conjunction with new analytical and interpretative methods, to demonstrate proof-ofprinciple capabilities of DCPM to measure colloidal interactions at the surface of live cells.23 In the present study, we now apply DCPM to measure nonspecific and specific interactions between live MDA231 epithelial breast cancer cells and a set of synthetic and biological macromolecules relevant to bioengineering applications, including polyethylene glycol (PEG), bovine serum albumin (BSA), dextran, and hyaluronic acid (HA) (Figure 1). By using micron-sized colloidal probes above the minimum size required for passive or active uptake into the cell,24−26 our study focuses on particles interacting with cell surfaces.
In the development of advanced biotechnologies, precise design of biomaterials (e.g., nanoparticle drug carriers, biosensors, tissue scaffolds) is required to control interactions at the surface of cells and tissues important to fundamental biological processes (e.g., protein adsorption, cellular uptake, cell adhesion).1−5 While many synthetic and biological materials have proven to be very useful in biomedical applications, a fundamental understanding of their interactions with live biological interfaces could enable the optimal design and implementation of such materials in new therapeutic and diagnostic tools. To develop models to better describe the interactions in these complex biological systems, highly sensitive, quantitative measurements are needed that are capable of directly interrogating the surfaces of cells and tissues. Several methods have been used to assess how the physicochemical properties of biomaterials and cells impact interactions at their interface. Spectroscopy-based imaging techniques, such as surface plasmon resonance and total internal reflection microscopy, have been used to measure the distribution of biomacromolecules at interfaces between cells and polymer, protein, and carbohydrate functionalized substrates.6−12 The interactions between these biomaterials and cells are generally inferred indirectly from changes in concentration of membrane-associated biomolecules (e.g., membrane protein clustering, actin stress fibers) and cell morphology (e.g., cell spreading, focal adhesions). Atomic force microscopy has been used to directly measure interactions of cells with biofunctionalized colloids and substrates.13−18 The deflection of a mechanical cantilever in contact with the surface of the cell is used to measure interfacial interactions and can also be used to determine the topography of cells and tissues. © XXXX American Chemical Society
Received: September 7, 2016 Revised: October 26, 2016 Published: October 27, 2016 A
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
interactions), and nonconservative forces (e.g., dissipative hydrodynamic interactions) with the underlying substrate or cell surface. At physiological ionic strengths where electrostatic interactions are short-ranged, steric interactions between macromolecular layers on the particle and surface are required to prevent irreversible adsorption of the particle, allowing it to diffuse laterally (and without tangential forces). While the macromolecular coating on the particles varied across cases, the silica substrates used in this work were coated with an artificial ECM coating of 100 kDa HA to provide both steric stability for the particles and an adherent surface for cell seeding. Low particle concentrations, with area fractions from 0.01−0.05, were used to limit particle−particle interactions that could affect particle trajectories and/or particle-cell surface interactions via multibody interactions.28 Dark field video microscopy was used for label-free imaging of colloids and live cells. Image analysis techniques were developed in prior work for automated detection of colloids and cell perimeters, allowing real-time, 2D multiparticle and multicell tracking (see ref 23 for a full description of the image analysis method).23 Figure 2A−D shows example images where the identity and location of particles and cells have been determined with cell boundaries marked with white lines and particle centers marked with red circles. Particle centers and cell boundaries are found in each image, and the trajectories of particles were determined allowing for measurement of particle dynamics both off the cell surface (red lines), and on the cell surface (green lines). The trajectories of the particles are highly dependent on their interactions with the underlying surface, and with appropriate analysis, particle dynamics can be measured on the cell surface and the background substrate. Hydrodynamic particle-surface interactions hinder the lateral motion of colloidal particles to rates significantly slower than its predicted Stokes−Einstein diffusivity, D0, (∼30% D0 for rolling in the absence of tangential forces on smooth surfaces).29−31 In addition, hydrodynamic interactions between macromolecules on the particle and surface will further influence the degree of hindrance to particle diffusion. The density, structure, and overall thickness of the macromolecular layers on each surface will control the layer permeabilites and short-range lubrication forces.29,32 Based on prior studies, the layer thickness of BSA,21 PEG,22 dextran,22 and HA33 coatings are 5, 10, 25, and 50 nm, respectively. Finally, local binding events through specific interactions will also ultimately determine the lateral particle motion on cells and macromolecule coated substrates. In prior work, total internal reflection microscopy measurements of kTscale BSA-BSA, PEG-BSA, and dextran-BSA interactions were found to be net repulsive.21,22 In the case of HA, we also expect net repulsive interactions with serum proteins since both have a net negative charge. Therefore, we do not expect significant adsorption of serum proteins to the surface of the biomaterialcoated colloids studied in this work. The mean squared displacement (MSD) was determined off the cell surface (on HA-coated substrate) and on the cell surface shown in Figure 2E and 2F, respectively, for each macromolecular coating. An average lateral diffusion coefficient, ⟨D∥⟩, was calculated from MSD data in Figure 2, and the mobility coefficient, f∥ = ⟨D∥⟩/D0, is shown for each case in Table 1. The MSDs of particles on the HA-coated substrate are shown in Figure 2E for each coating and show that the particles experience hindered diffusion due to hydrodynamic interactions on the background HA-coated surface. The measured diffusion rates on the background substrate ranged from ∼20−40% of
Figure 1. Schematic of colloidal probes fuctionalized with (A) pluronic-F108 copolymer (PEG) as a model synthetic antibiofouling coating, (B) bovine serum albumin (BSA) as a model serum protein corona, (C) dextran as a common tissue engineering scaffold, and (D) hyaluronic acid (HA) as a widely used tumor-targeting ligand (not to scale). Reported layer thicknesses based on prior measurements for PEG,22, BSA,21 dextran,22 and HA33 coatings produced by adsorption onto octadecanol (C 18 ) and/or chemisorption onto (3glycidyloxypropyl)trimethoxysilane (GPTMS)- and 3-aminopropyl)triethoxysilane (APTES)-modified substrates.
Dynamic and statistical mechanical analyses of colloidal trajectories on the background substrate and on the cell surface are used to determine local diffusivities, association lifetimes, and particle-cell surface interaction potentials. Ultimately, the ability to directly and nonintrusively measure nonspecific and specific interactions between diffusing colloidal probes with different biomacromolecular species at their surfaces and live cancer cell surfaces provides fundamental information useful for interpreting and designing such interactions into material systems for biomedical applications.
■
RESULTS AND DISCUSSION Particle Diffusion on Cells and Artificial ECM Substrate Surfaces. The trajectories of 2 μm diameter SiO2 colloidal probes functionalized with PEG, BSA, dextran, and HA, over an HA-coated silica substrate with adherent MDA231 epithelial breast cancer cells were measured and analyzed to assess nonspecific and specific biomacromolecular interactions. MDA231 cells were used in this study given their abundant surface expression of CD44, a known receptor for HA.27 The buoyant weight of micron-sized particles confines their trajectories to separations near the surface. Their lateral motion is highly sensitive to interactions corresponding to both conservative forces (e.g., electrostatic, van der Waals, steric B
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
interactions with the synthetic and biological macromolecules on the particle surface. Net repulsive steric interactions lead to hindered diffusion rates (determined by hydrodynamic interactions) on the cell surface for PEG, BSA, and dextran coated colloids. In comparison to diffusion on the background substrate, the MSD of PEG, BSA, and dextran coated particles showed a slight curvature over time on the surface of the cell. This is likely a result of colloidal migration due to cell surface topography, but the effect was limited for particles of this size as prior work showed they will be confined to lower elevations on the cell perimeter.23 A large reduction in diffusion rate ( f∥ = 0.02) was observed for particles coated with HA when diffusing on the cell surface. The MDA231 cell line used in this work has been thoroughly characterized and has shown abnormal overexpression of the transmembrane protein CD44.27 CD44 is known to interact specifically with HA,34 an extracellular matrix component widely used in drug delivery applications to specifically target cancer cells.35−38 Unlike the other biomaterial coatings, attractive particle−surface interactions are induced by specific CD44-HA interactions which lead to association of particles to the cell surface. While we can infer changes from net repulsive to net attractive particle−cell surface interactions based on measured lateral diffusion rate, the next sections will address how more information about colloidal interactions with the surface of the cell can be extracted from optical microscopy measurements of particle trajectories. Particle−Cell Surface Association Lifetimes. Compared to averaging over the whole trajectory as in the previous section, analysis of individual particle trajectories can provide local information on dynamics of particles interacting with cells that can be related to net particle−cell surface interactions. Mobile colloidal probes unassociated with the cell surface will progressively sample positions on the cell surface as depicted in Figure 3A. As a result, the standard deviation of x- and ypositions of colloidal probes, σxy, will vary over time depending on its lateral diffusion coefficient. Colloidal probes are defined as associated with the cell surface when σxy over time is less than a characteristic length scale La, calculated based on the theoretical diffusion-limited motion of a particle at contact with the cell surface.18,19,22 In Figure 3B, representative traces of σxy over time show PEG (black) and BSA (blue) functionalized colloidal probes progressively sample positions on the cell surface with σxy > La = 120 nm with La indicated as a dashed line in Figure 3B. However, for dextran and HA, association events of probe particles on the cell surface are more frequently detected with σxy < La as shown in green and red in Figure 3B, respectively. To quantitatively analyze association events to determine effective particle-cell surface interactions, the association lifetime of particles to the cell surface, ta, can be compared to a characteristic time scale for purely diffusive particles. In the limit of no net attraction with the cell surface, ta reduces to the diffusion-limited time τa and increases exponentially with only small increases in net particle−cell surface attraction.21−23 With this measurement, weak, intermittent association as well as strong, irreversible association to the cell surface can be detected with kT-scale sensitivity. Figure 3C−F shows measured association lifetime histograms, p(ta), for PEG, BSA, dextran, and HA-coated particles on the surface of MDA231 cells. As ta depends exponentially on the depth of the particle-cell surface attractive energy well, umin, histograms were populated from τa (6.14 s) to the total time of each experiment
Figure 2. (A−D) Colloidal trajectories of (A) PEG-coated, (B) BSAcoated, (C) dextran-coated, and (D) HA-coated colloidal silica particles (2 μm) interacting with MDA231 cells from DFVM experiments processed with image analysis. Cell boundaries at t = 0 are drawn as solid white lines. Particle centers are marked with a red circle, and trajectories are drawn as solid lines with their color depending on whether the particles are within the cell boundaries (green) and outside the cell boundaries (red). Green (i.e., on cell surface) particle trajectories appearing outside of cell boundaries are due to cell migration and the resulting cell boundary updates. (E,F) 1D lateral mean squared displacement averaged over the x and ydirection of PEG (circles), BSA (triangle down), dextran (squares), and HA (diamonds) coated 2 μm silica when (E) off MDA231 cell surface (red) and (F) on MDA231 cell surface (green). Dashed lines from theoretical fits to diffusion data used to determine ⟨D∥⟩. Trajectories included in the MSD analysis were on the order of 100 s on the substrate and 30 s on the cell surface with >250 trajectories for each case.
Table 1. Mobility Coefficient, f∥ = ⟨D∥⟩/D0, Determined from MSD Data (Figure 2) of Particles on the Background Substrate (HA) and on the Cell Surface (MDA231)a particle PEG BSA dextran HA
f∥, HA
f∥, MDA231
± ± ± ±
0.39 ± 0.03 0.31 ± 0.02 0.26 ± 0.05 0.02 ± 0.001
0.39 0.32 0.28 0.22
0.01 0.03 0.02 0.02
Data represents the average ± standard deviation of three independent measurements for each case. a
their predicted Stokes−Einstein diffusion rate (i.e., for an isolated particle at infinite separation from a surface). This gives us confidence when analyzing their trajectories on cell surfaces that particle diffusion can be treated as quasi-2D near the surface. MSDs of colloidal particles with different coatings on the MDA231 cell surface are shown in Figure 2F. A variety of membrane-associated biomolecules are present on the surface of the cell which can lead to nonspecific and specific C
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 3. (A) Illustration of unassociated and associated colloidal probes determined based on the standard deviation in x- and y- coordinates, σxy, of particle centers over time (not to scale). (B) Representative trajectories of σxy for PEG-coated (black), BSA-coated (blue), dextran-coated (green), and HA-coated (red) 2 μm silica particles on the surface of MDA231 cells. Dashed line indicates characteristic length scale (La) for discrimination of association events. (C−F) Association lifetime histograms, p(ta), for (C) PEG-coated, (D) BSA-coated, (E) dextran-coated, and (F) HA-coated 2 μm silica particles on the surface of MDA231 cells. Each bar has a linear color scale of ln(ta/τa) ≈ |umin|/kT shown as an inset in panel F.
case. In comparison to PEG, dextran-coated particles clearly shows longer lived association to the cell surface and weak, net attraction to the surface of the cell. Dextran-coated interfaces and hydrogels are widely used in antifouling and tissue engineering applications as an alternative to PEG to limit protein binding and cell adhesion.41,42 Our results suggest that weak, kT-scale interactions may exist between dextran and the surface of cells which may not be resolved with traditional methods such as cell attachment and proliferation assays. A potential explanation of dextran-mediated, weak attractive interactions with MDA231 cells are due to the carbohydratebinding galectins known to be expressed on their surface.43,44 The ensemble average lateral diffusion measured for dextran on the cell would indicate little change in particle dynamics compared to the background substrate. However, measured MSD would likely be insensitive to weak, reversible particle− surface association to the cell, producing only a small reduction in overall ensemble average diffusion rate. The association lifetime histogram for HA-coated particles, shown in Figure 3F, contrasted greatly with the short association lifetimes measured in other cases. A striking increase in long association lifetimes is observed for HA-coated particles with the mode at ta = 21.7 s (∼4τa). As overexpression of CD44 is common for MDA-231 and other cancerous cell types, CD44 binding of HA leads to increases in ta due to strong, net particle-cell surface interactions. While specific CD44-HA interactions are typically weak with dissociation constants, KD, ranging from 1 μM to 1 mM,45 multivalency on the particle scale will allow for many CD44-HA bonds to form in parallel and induce strong adhesion to the cell surface.46,47 Time-Averaged Quasi-Equilibrium Particle−Cell Surface Interactions. In contrast to the dynamic measures of local cell-surface interactions previously discussed, a statistical mechanical analysis can be used to assess quasi-equilibrium particle−cell interactions by analyzing time-averaged particle radial distributions with respect to the cell surface. The radial distance between particle centers and nearest cell boundary, r, can be measured, illustrated in Figure 4A, and histograms of
(1800 s) over 10 bins with exponential spacing. Each bin of p(ta) is normalized by the total number of measured association lifetimes so that the sum over bins is equal to 1. A linear color scale of |umin|/kT ≈ ln(ta/τa) is used for each bar in the association lifetime histograms shown in Figures 3C−F. A histogram of association lifetimes for PEG-coated particles to MDA231 cells, shown in Figure 3C, showed only minimal sampling of association lifetimes greater than τa. Particles primarily show short association times indicative of diffusion limited motion in the presence of net repulsive interactions with the cell surface. This is consistent with the average ensemble diffusion of PEG-coated particles on MDA231 cells, where particles were able to freely diffuse on the cell surface. The association lifetimes for BSA-coated particles, shown in Figure 3D, was similar to the PEG-coated particles with the mode association lifetime at τa and again was consistent with their ensemble average lateral diffusion. As serum adsorption is known to facilitate adhesion for in vitro cell culture and cellular uptake of nanoparticles, longer association lifetimes could be expected in this case. However, changes in particle-cell interactions have been observed for particles with preformed, adsorbed serum layers (as in this work) compared to serum from surrounding medium dynamically forming layers on the particle surface.39,40 The composition (i.e., profile of adsorbed serum proteins) and architecture (i.e., dense and irreversible versus soft and exchangeable serum layers) of the corona also greatly impact protein corona-mediated particle−cell interactions. As a result, the presence of serum proteins in the cell culture media could alter interactions of BSA-coated colloids with the cell surface via exchange with the preformed BSA protein corona. Future measurements with BSA-coated colloidal probes in serum-free media as well as uncoated colloidal probes in serum-containing media, thus allowing for dynamic protein corona formation, will be conducted to investigate these potential effects. While primarily short association lifetimes were measured for dextran-coated particles shown in Figure 3E, some weak, intermittent association lifetimes were also observed in this D
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 4. Time-averaged quasi-equilibrium distribution of colloidal probes on the surface of cells. (A) Illustration of measured particle center to nearest cell boundary point distance, r. Cell boundaries are drawn as solid white lines. Particle centers are marked with a red circle. Measured r are shown as dashed green lines. (B−E) Ensemble average particle-cell surface distribution function, p(r), for (B) PEG-coated, (C) BSA-coated, (D) dextran-coated, and (E) HA-coated 2 μm silica particles interacting with MDA231 cells.
Figure 5. (A) Schematic of biomaterial-coated colloidal probes interacting with an HA-coated interface with adherent MDA231 cells (not to scale). (B−E) Ensemble average particle-cell surface potentials of mean force, W(r), determined from distribution functions in Figure 4 for (B) PEG-coated, (C) BSA-coated, (D) dextran-coated, and (E) HA-coated 2 μm silica particles interacting with MDA231 cells. Solid lines are theoretical fits to measured potentials based on eqs 1−3
particle-cell surface separations, p(r), are constructed. Particle− cell surface separation histograms are shown for each case in Figure 4B−E. Changes in sampled r measured in the particle− cell radial distributions can be related to particle−cell surface interaction potentials. For the quasi-2D interactions of particles with cells in these experiments, this reduction in dimensionality provides a relevant measure of interactions between colloidal probes and the cell surface. Figure 5 shows the measured particle−cell surface interaction potentials for each particle coating. A reference state is chosen at r = 10a, where a is the particle radius, as it is assumed that no net interactions are present between the particle and the cell surface at this relatively large particle-cell distance. In the p(r) for PEG-coated particles shown in Figure 4B, a relatively constant distribution of distances are sampled at separations greater than 2a. At these separations far from the cell, the sampling will simply depend on the area fraction of particles on the surface which is observed with p(r) ≈ 1. Values of r < 0 are given for radial positions sampled within the cell boundaries and sampling at these positions remains relatively constant for PEG coated particles until separations in the range r < −2a. Decreased sampling at r < −2a was a common feature in the distributions across all cases. The p(r) for BSA (Figure 4C) shared similar features as described for PEG. The p(r) for dextran-coated particles (Figure 4D) differed from the PEG and BSA cases with increased sampling as the particle approaches the cell surface (r < 2a). HA-coated particles also showed an increase in sampling as they approached the cell surface (r < 2a) and even further increased sampling within the cell boundaries (−2a < r < 0) as shown in Figure 4E. The interactions of particles with the cell surface as a function of r, as illustrated in Figure 5A, will be dependent on
particle−cell surface biomolecular interactions and cell surface topography. Particle−cell surface interaction potentials, shown in Figure 5, were quantitatively determined from the measured p(r) shown in Figures 5B−E. From the particle−cell surface interaction potential for PEG-coated particles shown in Figure 5B, no net interactions are observed at separations greater than 2a, as would be expected at such large distances from the cell. As the particle approaches the surface (r < 2a), the free energy fluctuates around ∼0 kT with no appreciable attraction to the cell surface. As it diffused further onto the cell (r < −2a), a net repulsive interaction is seen. While no net surface interactions are expected based on measured association lifetimes for PEGcoated particles shown in Figure 3C, the presence of a gravitational barrier to diffusion laterally onto the cell due to cell topography (Figure 5B), as was found in our prior work,23 will lead to net repulsion as colloids approach the cell surface. A similar net repulsive potential was measured for BSA-coated particles (Figure 5C) with a potential fluctuating around 0 kT, also consistent with association lifetime measurements, before reaching the topographical barrier to positions on the top of the cell. Interaction potentials measured for dextran coated particles with MDA231 cells both showed net attraction at particle−cell surface contact (at r ≈ a) with repulsion to radial positions further on top of the cell (r < −2a) as shown in Figure 5D. With attractive minimum of ∼1−2 kT for dextran coated particles, reversible association to the cell surface would be expected and is consistent with measured association dynamics in Figure 4E. In Figure 5E, net attraction to the cell surface is E
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Langmuir
■
CONCLUSIONS Diffusing colloidal probes of biomaterial-cell surface interactions were used to assess natural and synthetic polymer coatings like dextran and PEG used in tissue engineering and drug delivery applications. In addition, the impact of a serum protein corona and cancer cell targeting surface moieties on particle−cell interactions was explored using colloidal probes functionalized with BSA and HA, respectively. Based on measured MSD, colloidal probes functionalized with PEG, BSA, and dextran displayed diffusion rates on the cell surface similar to that measured on the background HA-coated substrate. By contrast, the diffusion rate of HA-coated particles on the cell surface was significantly reduced compared to its transport rate on the background substrate. Measured association lifetimes and net particle−surface interaction potentials showed net repulsive interactions for PEG and BSA-coated colloids on the cell surface while dextran-coated colloids were weakly attractive to the cell surface. As HA-coated colloidal probes came within range of the cell surface, net attractive interactions to the cell surface were seen due to CD44-HA binding. Based on empirical fits, quasi-2D particle− cell surface interactions are well described with a net potential consisting of biomaterial-dependent steric and specific interaction potential combined with a cell topography-dependent gravitational potential. Taken together, the results of this work demonstrate a nonintrusive method to assess the interactions of biomaterials with live cells. In future work, nanoparticle probes capable of being internalized by cells will allow for further investigation into intracellular properties such as microviscosity and transport mechanisms with trajectories fit using the Fokker−Planck equation49 to obtain combined energy and diffusion landscapes (i.e., W(r) and D(r)).29 Ultimately, our findings establish a powerful technique for characterization of biomaterial−cell surface interactions crucial to biosensing, drug delivery, diagnostic imaging, and tissue engineering applications.
seen for HA-coated particles at radial positions on the cell (at −2a < r < a) due to the high expression of CD44 receptors on the MDA231 cell surface (Figure 5A). The attractive minimum of ∼4 kT will lead to much stronger particle cell−surface association and is again consistent with the long association lifetimes of HA-coated particles on the cell surface shown in Figure 4E. W(r) can be used as a composite measure of biophysical and biomolecular properties of the cell as differences in these potentials arise from the net contributions of biomacromolecular interactions and ensemble average topography of the cells. To consider each of these effects, particle−surface interaction potentials in Figure 5 were empirically fit with a theoretical net potential (eqs 1−3), which consists of a gravitational component, to account for cell topography, and a biomacromolecular component, to account for biomaterial mediated particle-cell surface interactions. Parameters used in these fits are summarized in Table 2 and a description of each Table 2. Parameters Used for Theoretical Fits to Particle− Cell Surface Potentials in Figure 5a parameter
PEG
BSA
dextran
HA
κ−1/μm umin/kT rmin/μm δ/μm ks−1/μm SE
24 0 2 8 1.6 0.13
18 0 2 8 1.6 0.34
24 3 2 8 1.6 0.02
30 6 2.5 8 1.3 0.02
Article
a The standard error (SE) of the theoretical fits was calculated for each case.
parameter can be found in the Methods section. The adjustable parameter κ relates to the steepness of repulsion due to cell topography. The steepness of repulsion varies from case to case most likely due to varying ensemble average topography, but the overall trend is consistent with a gravitational barrier to diffusion on top of the cell. For PEG and BSA-coated particles, no attractive energy well was present at short ranges (r < 2a), but dextran and HAcoated particles both showed attractive energy wells in their potential profiles. The fit parameters rmin and umin relate to the location and depth of the attractive energy well, respectively. The position of rmin was around −2a for both cases. However, for HA-coated colloids, umin was twice as large (6 kT vs 3 kT) as that found for dextran-coated colloids. This difference we can attribute to specific interactions between HA and CD44 versus the nonspecific interactions of dextran with the cell surface, which we would expect to be weaker. The umin in the theoretical fits are larger in magnitude than the measured well depth for each case, with ∼4 kT measured vs 6 kT fit for HA and ∼2 kT measured versus 3 kT fit for dextran. However, the net well depth results from the sum total of topographical repulsion and biomaterial-mediated attraction at the cell surface. This further illustrates the combined effects of biophysical and biomolecular properties on measured particle−cell surface interactions. In future work, recently developed computational models48 will be used to explicitly account for biomolecular interactions to understand their impact on particle−cell surface potentials to enhance the interpretive and predictive capabilities of our experimental approach.
■
METHODS
Preparation of PEG-, BSA-, Dextran-, and HA-Modified Colloidal Silica. Hydrophobically modified colloidal 2 μm diameter silica particles (Bangs Laboratories) were prepared with adsorbed layers of F108-Pluronic (PEG) and BSA using a procedure described in prior studies.22,50 Silica particles modified with an epoxysilane linker were used to attach dextran to their surface as described in prior studies.51 This protocol was then adapted for conjugation of HA using an amino-silane linker.33 Briefly, the particles were dispersed in a 2% (w/w) (3-aminopropyl)triethoxysilane (APTES, Sigma) in dry ethanol for 24 h at room temperature. The particles were then washed five times in dry ethanol and five times in deionized (DI) water. To conjugate hyaluronic acid (HA, 1 MDa, R&D Systems) to silica particles, amino-silane (APTES) modified colloids were dispersed in 3 mg/mL HA solutions for at least 20 h. The polysaccharide-modified particles are then centrifuged at 10 000 rpm for 10 min and redispersed in fresh DI water. They are then washed with DI water an additional five times. Pluronic-F108 (BASF) is then physisorbed onto the polysaccharide-modified particles by dispersing the particles in 1000 ppm (1 mg/mL) aqueous solution of F108-Pluronic overnight. The F108 coating step is to ensure the particles are fully coated with a polymer brush to improve colloidal stability. The particles are then rinsed five times in DI water and then dispersed in PBS. HA-Coated Coverslips with Adherent Cells. A protocol adapted from the literature was used to covalently attach HA to silica surfaces.33 Glass coverslips (18 mm × 18 mm, Corning Life Science) were first cleaned with sonication in acetone for 30 min and allowed F
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
where u0 = uG(r = 0) = 0. The form of this potential was chosen with the assumption that off the cell (r > 0) that topography will be unchanged and as the colloid approaches the cell surface (r ≤ 0), the cell height will steadily increase as the colloid moves more centrally toward the cell nucleus. The biomolecular potential accounts for nonspecific and specific interactions between a macromolecule-coated colloid and cell surface, and it is represented with an attractive Morse potential defined as
soak overnight in Nochromix. Next, they were rinsed 20 times with DI water followed by sonication in 0.1 M KOH for 30 min. They were again rinsed 20 times with DI water and dried with nitrogen. The coverslips were allowed to dry for an additional 30 min before being placed in 2% (w/w) APTES solution in ethanol for 24 h for aminosilane functionalization. To rinse excess APTES away from the surface, slides are first sonicated in ethanol for 30 min, followed by sonication in water for 30 min, and then dried with nitrogen. A 3 mg/mL HA (100 kDa, Lifecore Biomedical) solution was made in DI water filtered with an Anotop 0.02 μm syringe filter (Whatman) to ensure sterility. Five hundred microliters of the HA solution was placed on each APTES-coated coverslip and was kept covered in a dish coated with moist towels to chemisorb for at least 20 h. Once the HA chemisorption step was completed, the HA-coated coverslips were placed under UV irradiation for 30 min to ensure sterility and then placed into a 6-well plate (Corning Life Science) in PBS. MDA-MB231 epithelial breast cancer cells (MDA231, National Cancer Institute Physical Sciences-Oncology Center) were maintained in DMEM (Invitrogen) containing 10% (v/v) fetal bovine serum (FBS, Atlanta Biologicals).27 MDA231 cells were seeded onto HA-coated coverslips at a 1:4 ratio (∼50 000 cells/cm2) in complete media (10% FBS in DMEM). The cells are then allowed to adhere and spread onto the HA-coated coverslips overnight before each experiment. Batch Cell Assembly. To create batch cells for experiments, an Oring was adhered to a hydrophobic glass coverslip and 100 μL of 1000 ppm F108 in PBS was absorbed for ∼4 h. This F108 adsorption step was done to prevent nonspecific adhesion of colloidal particles to the coverslip. Excess, unadsorbed F108 was rinsed out and 100 μL of silica particles in complete media were added to the O-ring before being irradiated with UV for 30 min to sterilize the sample. The O-ring was then removed, and the smaller cell-seeded glass coverslip was placed on top of the large slide with particles. The entire assembly was then sealed with nail polish. The coverslip with F108 adsorbed is roughly 1 mm away from the cell-seeded coverslip after assembly and should not affect the measurements. The batch cell was then imaged using dark field video microscopy (DFVM) as described previously.23 Theoretical and Statistical Analyses. The image analysis algorithm as well as theoretical and statistical analyses used in this work are described in detail in prior studies.21−23 Briefly, the average lateral diffusion coefficient, defined as ⟨D∥⟩ = 0.5Δ⟨r2⟩/Δt, is determined from fitting the slope of measured MSD data. The mobility coefficient, defined as f∥ = ⟨D∥⟩/D0, is the mobility of the particle as compared to its Stokes−Einstein diffusivity, D0 = kT/6πμa, where kT is thermal energy, μ is the fluid viscosity, and a is the particle radius. For association lifetime analysis, the position of the particle was monitored for a characteristic diffusion-limited time, τa = 6.14 s (six consecutive images), and if the coordinates of in these images have a standard deviation σxy < La = 120 nm, the particle is considered to be associated with the surface.23 A colloid-surface attractive energy minimum, defined as |umin|/kT ≈ ln(ta/τa), is calculated from measured association lifetimes, ta. The particle−cell interaction potential, defined as W(r) = −ln[p(r)/p(rref)], is calculated via a Boltzmann inversion of measured particle−cell surface radial distributions, p(r) and rref is a chosen reference r where W(r) = 0. Measured W(r) were fit empirically with a net particle−cell surface potential as the superposition of contributing potentials, defined as
uN(r ) = uG(r ) + uB(r )
2 ⎧ ⎪u min[1 − exp[ − b(r − rmin) ] − umin , r ≥ rmin uB⎨ ⎪ 0, r < rmin ⎩
where umin is the energy well minimum, rmin is the location of the energy minimum, and b = (δ/ks)−1 is a fitting constant where ks is the spring constant and δ is the range of the interaction. We have used this potential in previous studies to model colloidal interactions mediated by specific biomolecular interactions.48,51,52
■
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We acknowledge financial support by the National Science Foundation (CBET-0834125, CBET-1066254, CHE-1112335, and an IGERT traineeship).
■
REFERENCES
(1) Nel, A. E.; Mädler, L.; Velegol, D.; Xia, T.; Hoek, E. M. V.; Somasundaran, P.; Klaessig, F.; Castranova, V.; Thompson, M. Understanding biophysicochemical interactions at the nano-bio interface. Nat. Mater. 2009, 8, 543−557. (2) Yameen, B.; Choi, W. I.; Vilos, C.; Swami, A.; Shi, J.; Farokhzad, O. C. Insight into nanoparticle cellular uptake and intracellular targeting. J. Controlled Release 2014, 190, 485−499. (3) Albanese, A.; Tang, P. S.; Chan, W. C. The effect of nanoparticle size, shape, and surface chemistry on biological systems. Annu. Rev. Biomed. Eng. 2012, 14, 1−16. (4) Pelaz, B.; Charron, G.; Pfeiffer, C.; Zhao, Y.; De La Fuente, J. M.; Liang, X. J.; Parak, W. J.; Del Pino, P. Interfacing engineered nanoparticles with biological systems: anticipating adverse nano−bio interactions. Small 2013, 9 (9−10), 1573−1584. (5) Leckband, D.; Israelachvili, J. Intermolecular forces in biology. Q. Rev. Biophys. 2001, 34 (02), 105−267. (6) Giebel, K.; Bechinger, C.; Herminghaus, S.; Riedel, M.; Leiderer, P.; Weiland, U.; Bastmeyer, M. Imaging of cell/substrate contacts of living cells with surface plasmon resonance microscopy. Biophys. J. 1999, 76, 509−516. (7) Hoover, D. K.; Lee, E.-J.; Yousaf, M. N. Total internal reflection fluorescence microscopy of cell adhesion on patterned self-assembled monolayers on gold. Langmuir 2009, 25, 2563−2566. (8) Parsons, J. T.; Horwitz, A. R.; Schwartz, M. A. Cell adhesion: integrating cytoskeletal dynamics and cellular tension. Nat. Rev. Mol. Cell Biol. 2010, 11, 633−643. (9) Steyer, J. A.; Almers, W. A real-time view of life within 100 nm of the plasma membrane. Nat. Rev. Mol. Cell Biol. 2001, 2, 268−275. (10) van der Merwe, P. A.; Barclay, A. N. Analysis of cell-adhesion molecule interactions using surface plasmon resonance. Curr. Opin. Immunol. 1996, 8, 257−261. (11) Wang, W.; Wang, S.; Liu, Q.; Wu, J.; Tao, N. Mapping singlecell-substrate interactions by surface plasmon resonance microscopy. Langmuir 2012, 28, 13373−9. (12) Ziblat, R.; Lirtsman, V.; Davidov, D.; Aroeti, B. Infrared surface plasmon resonance: a novel tool for real time sensing of variations in living cells. Biophys. J. 2006, 90, 2592−2599.
(1)
⎪ ⎪
AUTHOR INFORMATION
Corresponding Author
where the subscripts refer to gravitational (G) and biomolecular interaction (B) potentials. The gravitational potential energy accounts for the effect of cell topography on positions sampled by micron-sized colloids on the cell surface which will depend on r as
⎧G exp[− κr 2], r ≤ 0 uB⎨ 0, r>r ⎩
(3)
(2)
where, G = 4/3πa (ρp−ρf)g, is the buoyant weight of the particle, a is the particle radius, ρp is the particle density, ρf is the fluid density, g is the acceleration due to gravity, κ is an adjustable fit parameter related to cell topography, and u0 is the reference energy at the cell boundary, 3
G
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir (13) Dörig, P.; Ossola, D.; Truong, A. M.; Graf, M.; Stauffer, F.; Vörös, J.; Zambelli, T. Exchangeable Colloidal AFM Probes for the Quantification of Irreversible and Long-Term Interactions. Biophys. J. 2013, 105, 463−72. (14) Iyer, S.; Woodworth, C. D.; Gaikwad, R. M.; Kievsky, Y. Y.; Sokolov, I. Towards nonspecific detection of malignant cervical cells with fluorescent silica beads. Small 2009, 5, 2277−84. (15) Müller, D. J.; Dufrêne, Y. F. Atomic force microscopy: a nanoscopic window on the cell surface. Trends Cell Biol. 2011, 21, 461−9. (16) Ounkomol, C.; Yamada, S.; Heinrich, V. Single-cell adhesion tests against functionalized microspheres arrayed on AFM cantilevers confirm heterophilic E- and N-cadherin binding. Biophys. J. 2010, 99, L100−2. (17) Raman, a.; Trigueros, S.; Cartagena, a.; Stevenson, a. P. Z.; Susilo, M.; Nauman, E.; Contera, S. A. Mapping nanomechanical properties of live cells using multi-harmonic atomic force microscopy. Nat. Nanotechnol. 2011, 6, 809−14. (18) Vasir, J. K.; Labhasetwar, V. Quantification of the force of nanoparticle-cell membrane interactions and its influence on intracellular trafficking of nanoparticles. Biomaterials 2008, 29, 4244−52. (19) Neuman, K. C.; Nagy, A. Single-molecule force spectroscopy: optical tweezers, magnetic tweezers and atomic force microscopy. Nat. Methods 2008, 5 (6), 491−505. (20) Everett, W. N.; Bevan, M. A. kT-Scale interactions between supported lipid bilayers. Soft Matter 2014, 10, 332. (21) Everett, W. N.; Wu, H.-j.; Anekal, S. G.; Sue, H.-j.; Bevan, M. A. Diffusing Colloidal Probes of Protein and Synthetic Macromolecule Interactions. Biophys. J. 2007, 92, 1005−1013. (22) Eichmann, S. L.; Meric, G.; Swavola, J. C.; Bevan, M. A. Diffusing colloidal probes of protein-carbohydrate interactions. Langmuir 2013, 29, 2299−2310. (23) Duncan, G. A.; Fairbrother, D. H.; Bevan, M. A. Diffusing Colloidal Probes of Cell Surfaces. Soft Matter 2016, 12, 4731−4738. (24) Cho, K.; Wang, X.; Nie, S.; Chen, Z. G.; Shin, D. M. Therapeutic nanoparticles for drug delivery in cancer. Clin. Cancer Res. 2008, 14, 1310−1316. (25) Mailänder, V.; Landfester, K. Interaction of nanoparticles with cells. Biomacromolecules 2009, 10, 2379−400. (26) Chauhan, V. P.; Jain, R. K. Strategies for advancing cancer nanomedicine. Nat. Mater. 2013, 12, 958−62. (27) Agus, D. B.; Alexander, J. F.; Arap, W.; Ashili, S.; Aslan, J. E.; Austin, R. H.; Backman, V.; Bethel, K. J.; Bonneau, R.; Chen, W.-C.; Chen-Tanyolac, C.; Choi, N. C.; Curley, S. A.; Dallas, M.; Damania, D.; Davies, P. C. W.; Decuzzi, P.; Dickinson, L.; Estevez-Salmeron, L.; Estrella, V.; Ferrari, M.; Fischbach, C.; Foo, J.; Fraley, S. I.; Frantz, C.; Fuhrmann, A.; Gascard, P.; Gatenby, R. A.; Geng, Y.; Gerecht, S.; Gillies, R. J.; Godin, B.; Grady, W. M.; Greenfield, A.; Hemphill, C.; Hempstead, B. L.; Hielscher, A.; Hillis, W. D.; Holland, E. C.; IbrahimHashim, A.; Jacks, T.; Johnson, R. H.; Joo, A.; Katz, J. E.; Kelbauskas, L.; Kesselman, C.; King, M. R.; Konstantopoulos, K.; Kraning-Rush, C. M.; Kuhn, P.; Kung, K.; Kwee, B.; Lakins, J. N.; Lambert, G.; Liao, D.; Licht, J. D.; Liphardt, J. T.; Liu, L.; Lloyd, M. C.; Lyubimova, A.; Mallick, P.; Marko, J.; McCarty, O. J. T.; Meldrum, D. R.; Michor, F.; Mumenthaler, S. M.; Nandakumar, V.; O’Halloran, T. V.; Oh, S.; Pasqualini, R.; Paszek, M. J.; Philips, K. G.; Poultney, C. S.; Rana, K.; Reinhart-King, C. A.; Ros, R.; Semenza, G. L.; Senechal, P.; Shuler, M. L.; Srinivasan, S.; Staunton, J. R.; Stypula, Y.; Subramanian, H.; Tlsty, T. D.; Tormoen, G. W.; Tseng, Y.; van Oudenaarden, A.; Verbridge, S. S.; Wan, J. C.; Weaver, V. M.; Widom, J.; Will, C.; Wirtz, D.; Wojtkowiak, J.; Wu, P.-H. A physical sciences network characterization of non-tumorigenic and metastatic cells. Sci. Rep. 2013, 3, 1449. (28) Anekal, S.; Bevan, M. A. Self diffusion in sub-monolayer colloidal fluids near a wall. J. Chem. Phys. 2006, 125, 034906. (29) Beltran-Villegas, D. J.; Edwards, T. D.; Bevan, M. A. Selfconsistent colloidal energy and diffusivity landscapes in macromolecular solutions. Langmuir 2013, 29, 12337−41.
(30) Brenner, H. The Slow Motion of a Sphere Through a Viscous Fluid Towards a Plane Surface. Chem. Eng. Sci. 1961, 16 (3−4), 242− 251. (31) Goldman, A. J.; Cox, R. G.; Brenner, H. Slow Viscous Motion of a Sphere Parallel to a Plane Wall – I. Motion Through a Quiescent fluid. Chem. Eng. Sci. 1967, 22, 637−651. (32) Potanin, A. A.; Russel, W. B. Hydrodynamic interaction of particles with grafted polymer brushes and applications to rheology of colloidal dispersions. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 1995, 52 (1), 730−737. (33) Das, S.; Banquy, X.; Zappone, B.; Greene, G. W.; Jay, G. D.; Israelachvili, J. N. Synergistic interactions between grafted hyaluronic acid and lubricin provide enhanced wear protection and lubrication. Biomacromolecules 2013, 14, 1669−77. (34) Zöller, M. CD44: can a cancer-initiating cell profit from an abundantly expressed molecule? Nat. Rev. Cancer 2011, 11, 254−67. (35) Arosio, D.; Manzoni, L.; Araldi, E. M. V.; Scolastico, C. Cyclic RGD functionalized gold nanoparticles for tumor targeting. Bioconjugate Chem. 2011, 22, 664−72. (36) Cho, H.-J.; Yoon, H. Y.; Koo, H.; Ko, S.-H.; Shim, J.-S.; Lee, J.H.; Kim, K.; Kwon, I. C.; Kim, D.-D. Self-assembled nanoparticles based on hyaluronic acid-ceramide (HA-CE) and Pluronic® for tumor-targeted delivery of docetaxel. Biomaterials 2011, 32, 7181−90. (37) Choi, K. Y.; Chung, H.; Min, K. H.; Yoon, H. Y.; Kim, K.; Park, J. H.; Kwon, I. C.; Jeong, S. Y. Self-assembled hyaluronic acid nanoparticles for active tumor targeting. Biomaterials 2010, 31, 106− 14. (38) Choi, K. Y.; Yoon, H. Y.; Kim, J.-H.; Bae, S. M.; Park, R.-W.; Kang, Y. M.; Kim, I.-S.; Kwon, I. C.; Choi, K.; Jeong, S. Y.; Kim, K.; Park, J. H. Smart nanocarrier based on PEGylated hyaluronic acid for cancer therapy. ACS Nano 2011, 5, 8591−9. (39) Monopoli, M. P.; Walczyk, D.; Campbell, A.; Elia, G.; Lynch, I.; Bombelli, F. B.; Dawson, K. A. Physical-chemical aspects of protein corona: relevance to in vitro and in vivo biological impacts of nanoparticles. J. Am. Chem. Soc. 2011, 133, 2525−34. (40) Jedlovszky-Hajdú, A.; Bombelli, F. B.; Monopoli, M. P.; Tombácz, E.; Dawson, K. A. Surface coatings shape the protein corona of SPIONs with relevance to their application in vivo. Langmuir 2012, 28, 14983−91. (41) Sun, G.; Shen, Y. I.; Ho, C. C.; Kusuma, S.; Gerecht, S. Functional groups affect physical and biological properties of dextranbased hydrogels. J. Biomed. Mater. Res., Part A 2010, 93 (3), 1080− 1090. (42) Massia, S. P.; Stark, J.; Letbetter, D. S. Surface-immobilized dextran limits cell adhesion and spreading. Biomaterials 2000, 21 (22), 2253−2261. (43) Zhang, H.; Liang, X.; Duan, C.; Liu, C.; Zhao, Z. Galectin-3 as a marker and potential therapeutic target in breast cancer. PLoS One 2014, 9 (9), e103482. (44) Nangia-Makker, P.; Conklin, J.; Hogan, V.; Raz, A. Carbohydrate-binding proteins in cancer, and their ligands as therapeutic agents. Trends Mol. Med. 2002, 8 (4), 187−192. (45) Banerji, S.; Wright, A. J.; Noble, M.; Mahoney, D. J.; Campbell, I. D.; Day, A. J.; Jackson, D. G. Structures of the Cd44-hyaluronan complex provide insight into a fundamental carbohydrate-protein interaction. Nat. Struct. Mol. Biol. 2007, 14, 234−9. (46) Dubacheva, G. V.; Curk, T.; Auzély-Velty, R.; Frenkel, D.; Richter, R. P. Designing multivalent probes for tunable superselective targeting. Proc. Natl. Acad. Sci. U. S. A. 2015, 112 (18), 5579−5584. (47) Santore, M. M.; Zhang, J.; Srivastava, S.; Rotello, V. M. Beyond molecular recognition: using a repulsive field to tune interfacial valency and binding specificity between adhesive surfaces. Langmuir 2009, 25 (1), 84−96. (48) Duncan, G. A.; Bevan, M. A. Computational design of nanoparticle drug delivery systems for selective targeting. Nanoscale 2015, 7, 15332−15340. (49) Beltran-Villegas, D. J.; Sehgal, R. M.; Maroudas, D.; Ford, D. M.; Bevan, M. A. Fokker−Planck analysis of separation dependent H
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir potentials and diffusion coefficients in simulated microscopy experiments. J. Chem. Phys. 2010, 132 (4), 044707. (50) van Helden, A. K.; Jansen, J. W.; Vrij, A. Preparation and Characterization of Spherical Monodisperse Silica Dispersions in NonAqueous Solvents. J. Colloid Interface Sci. 1981, 81 (2), 354−368. (51) Duncan, G. A.; Bevan, M. A. Tunable Aggregation by Competing Biomolecular Interactions. Langmuir 2014, 30, 15253− 15260. (52) Duncan, G. A.; Bevan, M. A. Colloidal potentials mediated by specific biomolecular interactions. Soft Matter 2014, 10, 8524−8532.
I
DOI: 10.1021/acs.langmuir.6b03302 Langmuir XXXX, XXX, XXX−XXX
