Single-Molecule Kinetics of Protein Adsorption on ... - ACS Publications

Sep 6, 2016 - Wenxiao Wang , Hao Shen , Bo Shuang , Benjamin S. Hoener ... Nicholas A. Moringo , Hao Shen , Logan D.C. Bishop , Wenxiao Wang ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/ac

Single-Molecule Kinetics of Protein Adsorption on Thin Nylon-6,6 Films Hao Shen,† Lawrence J. Tauzin,† Wenxiao Wang,‡ Benjamin Hoener,† Bo Shuang,† Lydia Kisley,† Anneli Hoggard,† and Christy F. Landes*,†,‡,§ †

Department of Chemistry, ‡Department of Electrical and Computer Engineering, and §Smalley-Curl Institute, Rice University, Houston, Texas 77251, United States S Supporting Information *

ABSTRACT: Understanding and controlling protein adsorption on surfaces is critical to a range of biological and materials applications. Kinetic details that provide the equilibrium and nonequilibrium mechanisms are difficult to acquire. In this work, single-molecule fluorescence microscopy was used to study the adsorption of Alexa 555 labeled α-lactalbumin (αLA) on two chemically identical but morphologically different polymer surfaces: flat and porous nylon-6,6 thin films. The adsorption kinetics of spatially resolved single molecule α-LA binding to nylon films were quantified by a monolayer adsorption model. The surface morphology of the porous nylon-6,6 films increased the number of adsorption sites but decreased the binding affinity compared to the flat films. Such single-molecule based kinetic studies may be extended to various proteinpolymer interactions.

T

example, enzyme catalysis,22 ion−surface interaction,23−25 protein folding and unfolding,26 protein−DNA interaction, and heterogeneous catalysis.27−29 Such real-time observations also facilitate the kinetic measurements of protein−polymer interactions. For example, the Schwartz group used this technique to study the interaction of fibrinogen with poly(ethylene glycol) (PEG)30 and poly(ethylene) (HDPE)31 as well as the adsorption of bovine serum albumin (BSA) and immunoglobulin G (IgG) to regenerated cellulose (RC) and poly(ether sulfone) (PES)32 surfaces. In this work, single-molecule fluorescence microscopy was used to study the kinetics of protein−polymer interactions between α-lactalbumin (α-LA) and thin nylon-6,6 films. Nylon6,6 was chosen because it is a widely used membrane material for protein filtrations.33,34 The surface of nylon-6,6 can be functionalized to have different charge and wettability,35,36 and it is stable under constant laser illumination.37 Single-molecule fluorescence microscopy provides a direct way to image the protein adsorption. The super-resolution imaging technique confirmed that the α-LA molecules were adsorbed to nonspecific surface sites on nylon-6,6. By monitoring the number of newly adsorbed α-LA per unit time, we directly titrated the rate of adsorption under three different α-LA concentrations. It was found that the adsorption of α-LA onto a flat nylon-6,6 film followed a monolayer adsorption model. With this model, we were able to quantitatively determine the rate constants of

he adsorption of proteins onto polymers is of great importance in developing biomedical devices,1,2 processing food and beverages,3,4 and separating proteins.5 However, studying protein−polymer interactions is experimentally difficult because both the adsorbate and adsorbent are macromolecules and exhibit heterogeneous structural, transport, and reaction dynamics on multiple scales. Factors that may affect the adsorption process include the charge, polarity, and surface morphology of the polymer as well as the orientation and possible conformational changes of the protein adsorbed.6−10 To quantify the adsorption process, one needs to determine the amount of protein adsorbed and the rate of adsorption. Traditionally, quantification is achieved by gravimetric methods,8,11 surface plasmon resonance,12 fluorescent labeling,13 and dynamic light scattering (DLS).14 Despite their usefulness, these methods have their drawbacks, for example, lacking sufficient time-resolution to observe the kinetics or ensemble averaging of the results. Single-molecule fluorescence microscopy is a powerful tool to study protein−surface interactions.15−21 Rapid technological advances have made it possible to detect a single protein molecule with a fluorescent label, allowing hundreds of protein molecules to be monitored individually to build up significant statistics. The relatively small evanescent field in total internal reflection fluorescence (TIRF) microscopy restricts the observation volume to the interface, which eliminates background signal from labeled protein molecules in bulk solution. Singlemolecule fluorescence microscopy also allows for real-time observation with millisecond resolution, which facilitates the direct kinetic measurements of various dynamic processes, for © XXXX American Chemical Society

Received: October 28, 2015 Accepted: September 6, 2016

A

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

each movie, only molecules within a 240 by 360 pixel (15.4 by 23.0 μm) area at the center of imaging chip were counted and retained for analysis. Fluorescent events that appeared at the same location over consecutive frames were counted as one single protein molecule, and the total number of adsorbed molecules was summed over the entire 1 000 frames. This number was divided by the underlying surface area (354 μm2) and recording time (30 s) to obtain the area normalized rate of adsorption. For the porous films, their projected surface areas were used. This tracking program is available at www.lrg.rice. edu/Content.aspx?id=96.

adsorption, and thereby estimate the free energies for the adsorption process. For comparison, the adsorption of α-LA on porous nylon-6,6 films was also studied. The morphology of the adsorbent was found to appreciably affect the adsorption kinetics. We envision that such single-molecule based measurements can facilitate the imaging and surface analysis in protein−polymer interactions.



EXPERIMENTAL SECTION Substrate Preparation. Borosilicate glass coverslips (22 mm × 22 mm, VWR) were sequentially sonicated for 15 min in water, ethanol, and acetone. The substrates were then immersed in 80 °C base piranha solution (28% NH4OH and 30% H2O2) for 20 min. The coverslips were then dried with compressed nitrogen and further treated with oxygen plasma for 3 min. Synthesis and Characterization of Nylon-6,6 Films. Thin nylon-6,6 films were spin coated on precleaned coverslips. Solutions of formic acid and formic acid/ethanol mixture (9:1 v/ v) containing 1.5 wt % nylon-6,6 were used for the preparation of flat and porous nylon-6,6 films, respectively. In each case, the solution was applied to the center of the slide and spin coated at 3,000 rpm for 60 s to yield ∼70 nm thick films. The morphologies of the nylon-6,6 films were examined by an SEM (FEI QUANTA 650), the pore sizes of porous nylon-6,6 films were measured by an AFM (Bruker Multimode 8), and the thickness of all films was measured by an ellipsometer (Gaetner Scientific, D LSE C370). Single-Molecule Fluorescence Microscopy. Single-molecule fluorescence measurements were performed using a custom-built TIRF microscope. A 10 mW continuous-wave circularly polarized 532 nm laser (Coherent, compass 315M100SL) was used to excite the fluorescent probes. The excitation light passed through a 100× NA 1.45 oil-immersion objective (alpha Plan-Fluar, Carl Zeiss) and was focused onto an area of ∼30 × 20 μm2 in a microfluidic chamber, which was assembled on top of the sample slide. The fluorescence signal from probe molecules was collected by the same objective, filtered by two filters (Kaiser, HNPF-532.0-1.0 and Chroma, ET585/65m), and projected onto an EMCCD camera (Andor, iXon 897) operated in frame transfer mode and controlled by Andor Solis software. The time resolution of image acquisition was 30 ms. The nylon6,6 films were thoroughly rinsed with pH 7.3 HEPES buffer (GIBCO) to remove formic acid residue and the imaging area was illuminated with the 532 nm laser for an extensive period of time to photobleach any fluorescent contaminants prior to single-molecule measurements. For the protein adsorption measurement, α-LA solution was diluted in 10 mM pH 7.3 HEPES buffer and supplied continuously at 10 μL/min by a syringe pump (Kent Scientific, Genie Plus) to provide a nonequilibrium, steady state environment. During the measurement, a sequence of 1 000 frame (30 s) movies was collected at the same location of the nylon-6,6 film. The time interval between two consecutive movies was ∼2 min, and the total length of recording time was longer than 1 h. Protein Molecule Identification and Tracking. A previously described Troika program was used to identify and track the movements of α-LA molecules.38 Briefly, this Matlabbased program performs three sequential steps: (1) increasing the signal-to-noise ratio by applying a 3 by 3 pixel averaging matrix, (2) identifying protein molecules by searching for the local intensity maximum and then refining the center positions of molecules through radial symmetry fitting,39 (3) mapping molecular trajectories in consecutive frames by applying a nearest neighbor algorithm to approach the global optimum. For



RESULTS AND DISCUSSION α-LA Adsorption on Flat Nylon-6,6. Adsorption events were monitored at the aqueous solution/nylon-6,6 film interface (Figure 1A). Thin nylon-6,6 films are transparent, with a

Figure 1. (A) Experimental design based on TIRF microscopy. (B) Histograms of integrated bead intensities on bare glass (top) and ∼200 nm nylon-6,6 film (bottom), respectively. The intensity of each bead is integrated within a 7 by 7 pixel area. Red curves are Gaussian fits, and the insets are schemes of the bead deposition.

refractive index of ∼1.58, close to that of glass, in comparison to the estimated solution refractive index of 1.33. Thus, the conditions of TIR are met at the solution/film interface where the difference in refractive index is greatest. To confirm this condition experimentally, we drop-cast fluorescent beads onto a bare glass slide and a ∼200 nm thick nylon-6,6 film (Figure 1B, insets). The intensity of each bead was integrated within a 7 by 7 pixel area and the intensity distributions are shown in Figure 1B. The average intensity for beads on the bare glass slide is (1.3 ± 0.6) × 104 counts, while those on the ∼200 nm thick nylon-6,6 film is (1.1 ± 0.5) × 104 counts. The similar intensity for these two cases confirms that the TIR takes place at the solution/film interface; otherwise, the beads on the ∼200 nm thick nylon film would be much dimmer due to the fast decay of the evanescent field (Figure S2 in the Supporting Information). With the aid of subdiffraction localization of protein adsorption at the solution/film interface, we observe that the α-LA molecules adsorb to nonspecific adsorption sites on the nylon-6,6 films (Figure 2). When Alexa 555-labeled α-LA was introduced into the flow cell, the probe molecules were carried by fast flow, whose motions were too fast to be captured by the EMCCD camera (30 ms per frame) due to motion blur. In contrast, α-LA adsorption events to the nylon-6,6 films appeared as individual bright spots over the background (Figure 2A). Each B

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. Subdiffraction mapping of α-LA adsorption to a flat nylon-6,6 film. (A) Fluorescence image of α-LA (2 nM in pH 7.3 HEPES buffer) binding onto a flat nylon-6,6 film. This image was taken with a 30 ms frame time. (B) Central localizations for one single α-LA molecule for 30 consecutive frames marked by the arrow in part A. (C) 2D histogram of central localization from 20 α-LA molecules combined. Fitting this histogram with a 2D Gaussian gives FWHM ∼20 nm. (D) Histogram of diffusion coefficients for 400 α-LA molecules. Red curve is a Gaussian fit.

× 104 nm2 s−1 (Figure 2D), which matches with the steady state of stationary molecules described in our previous work,18 indicating that the α-LA molecules are immobilized rather than diffusing near the film. Moreover, the α-LA and nylon-6,6 interactions are restricted to the solution/film interface, as evidenced by the fact that small organic dye molecules sandwiched within nylon-6,6 films were immobilized (Figure S3 in the Supporting Information). Adsorption Kinetics. The rate of α-LA adsorption to surface sites on flat nylon-6,6 films was found to be time and concentration dependent (Figure 3). Spin coating 1.5 wt % nylon-6,6 in formic acid solution yielded ∼70 nm thick homogeneous films (Figure 3A and Figure S1 in the Supporting Information). AFM measurements indicated that the surface roughness of the as-prepared film is ∼0.46 nm (Figure 3B). To determine the adsorption kinetics, three α-LA concentrations were studied (2 nM, 5 nM, and 10 nM). Single-event detection was the key to the accurate measurements for the rate of adsorption, since the rate is defined as the number of newly adsorbed molecules per second, which is further normalized by the projected surface area. Therefore, we purposefully set the highest α-LA concentration to 10 nM to decrease the possibility of multiple adsorption events at the same adsorption site. Further analysis on the shape of point spread functions of adsorbed protein molecules confirmed that the adsorption was taking place at the single-molecule level (Figure S4 in the Supporting Information). When α-LA solution was flowed over the nylon6,6 film, the rate of adsorption was found to decay over time and eventually flatten out. Also, the absolute adsorption rate increased with increasing α-LA concentration (Figure 3C). The initial decay of the adsorption rate was a result of surface sites being occupied by α-LA molecules, while the flattening toward later times indicated that adsorption equilibrium had been reached. These observations support a Langmuir

bright spot lasted from one frame up to a few hundreds of frames before it disappeared. The duration of the α-LA on the film indicates the on-time of an individual protein molecule. In order to determine whether each protein adsorbed at a single location or laterally diffused near the surface, we used subdiffraction localization to map out the central localization of each protein’s fluorescence during its on-time. Specifically, for a protein molecule detected on the nylon-6,6, each individual frame during an on-time period was fit to a 2D Gaussian function to determine the centroid position and static localization uncertainty. The compiled centroid localizations from all frames are within an area of tens of nanometers as shown in Figure 2B. Similar approaches have been used in previous subdiffraction methods such as photoactivation localization microscopy (PALM)40 and stochastic optical reconstruction microscopy (STORM).41 In order to increase the sample size for statistics, we combined the localizations of 20 α-LA molecules by aligning the center of mass of central localizations, according to previous work.42,43 The 2D histogram (Figure 2C) was fit to another Gaussian function, and the full width at half maximum (FWHM) was ∼20 nm, comparable to the resolution limit of other superresolution techniques of stationary molecules reported in previous work by us44 and others.40,45,46 Therefore, we can conclude that adsorbed α-LA molecules were immobilized within our subdiffraction resolution. Single particle tracking using a maximum-likelihood estimation (MLE)38 also further supports that the adsorbed α-LA molecules are immobilized at the nylon-6,6 surface (Figure 2D). Briefly, we tracked the single frame displacements of each probe molecule as it moved from one frame to the next and calculated the corresponding diffusion coefficient. Thus, the distribution of the diffusion coefficients is the measure of average displacements for many probe molecules. It is found that the log(D) distribution takes a Gaussian shape with an average value 1.16 C

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 3. α-LA adsorption onto flat nylon-6,6 films. (A) SEM of flat nylon-6,6 film and (B) AFM of flat nylon-6,6 film. The top plot is one line scan marked by the red line. (C) Dependence of the adsorption rate versus time. The α-LA concentration are 2 nM (blue), 5 nM (green), and 10 nM (red), respectively. Each data point represents the average rate of adsorption within a 30 s observation window; solid lines are global fits using eq 5, and error bars are the standard deviation. (D) ⟨τon⟩−1 under various α-LA concentrations. Each data bar is the inverse averaged on-time of all protein molecules under the same conditions. The error bars represent the standard error of the mean.

adsorption model in which α-LA molecules adsorb to surface sites: if the observed events were dominated by random hitting, i.e., collisions of protein molecules with the surface that do not result in adsorption, the number of observed events would be time independent. A control example of random hitting is rhodamine-labeled lysozyme adsorption onto nylon-6,6 films under the same conditions (Figure S5 in the Supporting Information). Lysozyme was chosen because it has similar size and secondary and tertiary structures as α-LA, and it exhibited the expected time-independent rate of adsorption. It is also worth noting that for all of the tested α-LA concentrations (2−10 nM), the number of α-LA molecules supplied per unit time was at least 1 order of magnitude larger than the amount adsorbed onto the nylon-6,6 films. Therefore, the α-LA concentration in bulk solution remained constant. We propose a kinetic model of monolayer adsorption that can account for the time and α-LA concentration dependence of adsorption rates. When the surface coverage of α-LA on nylon6,6 is low, the intramolecular interactions of adsorbed α-LAs are negligible and the rate of adsorption is proportional to the bulk solution concentration of α-LA. Therefore, the total effective number of adsorption sites under low surface coverage, Pmax (Scheme 1), and the rate of adsorption, rads, can then be expressed by the following equation,47 rads = kads[P](Pmax − Pads)

Scheme 1. Kinetic Scheme of α-LA Sorption onto Nylon 6,6 Filmsa

a

kads and kdes are the rate constants for adsorption and desorption, respectively.

where kads is the adsorption rate constant, [P] the concentration of α-LA in bulk solution, and Pads the current amount of adsorbed α-LA. The term (Pmax − Pads) stands for the number of unoccupied sites under low surface coverage. The rate of desorption is dependent on the amount of surface adsorbed α-LA and takes the form of

(1) D

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Table 1. Kinetic and Thermodynamic Parameters for the α-LA Adsorption on Nylon-6,6 Films kads nM−1 s−1 flat film porous film

(1.3 ± 0.5) × 10 (3 ± 1) × 10−4

kdes s−1 −3

Pmax molecule nm−2 −4

−5

(1.9 ± 0.2) × 10 (1.8 ± 0.1) × 10−4

rdes = kdesPads

(8 ± 4) × 10 (3 ± 1) × 10−4

where kdes is the desorption rate constant. The net adsorption rate of α-LA on the nylon-6,6 surface is (3)

Rearranging eq 3 and integrating over time, the amount of surface adsorbed α-LA takes the form Pads =

kads[P]Pmax {1 − e−(kads[P] + kdes)t } kdes + kads[P]

(4)

Replacing Pads in eq 1 with eq 4: rads = kads[P]Pmax − e

−(kads[P] + kdes)t

ΔGads kJ mol−1

7±2 1.7 ± 0.6

−4.7 ± 0.9 −1.3 ± 0.8

conditions represent convolutions between the first order protein−support interactions they are meant to convey, and other hidden, higher order interactions. This possibility can be explored in future experiments with faster detectors and/or better imaging analysis algorithms. The equilibrium constant Keq is high, because of the small desorption rate constant. The van der Waals forces as well as the hydrophobic interactions might be the driving forces for the α-LA and nylon interactions, while the existence of hydrogen donors and acceptors on the nylon surface might facilitate the formation of multiple hydrogen bonds with a single α-LA molecule. The electrostatic interaction may not be strong in this case, as confirmed by the control of electrostatic shielding (Figure S7 in the Supporting Information). The change of free energy during the adsorption is related to the equilibrium constant:

(2)

dPads = kads[P](Pmax − Pads) − kdesPads dt

Keq nM−1

kads 2[P]2 Pmax k 2[P]2 Pmax + ads kadsPmax + kdes kadsPmax + kdes

ΔGads = −RT ln Keq (5)

(6)

Herein, the ΔGads for α-LA adsorption onto flat nylon-6,6 equaled −4.7 ± 0.9 kJ mol−1. It should be noted that under certain circumstances, the singlemolecule dwell-time analysis also provides the kdes value. For instance, α-LA desorption is a pseudo-first order process that is independent of the [P] in bulk solution. If the disappearance of fluorescent events is due to the desorption of α-LA molecules, then ⟨τon⟩−1, the inverse average of the fluorescence on time, is quantitatively described by the following equation:22,53,54

Eq 5 is the expression for the rate of adsorption. It takes the form of a single-exponential decay which agrees well with our experimental observations (Figure 3C). Eq 5 also predicts that under extremely low protein concentration, the time-dependent rate of adsorption can no longer be observed because the terms containing higher order [P] values all vanish, leaving rads = kads[P] Pmax, which is time-independent. We globally fit the data in Figure 3C with eq 5. Because of the inevitable random hitting events in our flow system discussed earlier, the fitting curve for the lowest [P] (blue and green curves in Figure 3C) deviated slightly from the experimental data, as fewer α-LA molecules adsorbed to nylon-6,6 under such conditions. However, the global fit showed that with higher [P], the fitted curves agreed well with the experimental results (Figure S6 in the Supporting Information). It was found that for flat nylon-6,6 films kads, kdes, and Pmax were (1.3 ± 0.5) × 10−3 nM−1 s−1, (1.9 ± 0.2) × 10−4 s−1, and (8 ± 4) × 10−5 molecule nm−2, respectively (Table 1). The equilibrium constant Keq, which is equivalent to the ratio of kads and kdes, was found to be 7 ± 2 nM−1 for flat nylon-6,6 films. The small Pmax is consistent with nylon-6,6 being hydrophilic, which leads to smaller loading capacity of soluble proteins.51,52 However, our calculated Pmax value is even smaller than other reported loading capacities acquired through ensemble methods.48−50 One explanation for this discrepancy is that, because of the higher protein concentrations necessary for ensemble analyses, protein aggregation and/or multiple protein adsorption contribute to overcounting in ensemble measurements. Single-molecule studies do not suffer from such effects. Alternatively, but less likely, there might be faster adsorption events that are not detectable at our camera frame rates, resulting in undercounting. Our experimental data suggest that any such undercounting, if it exists, contributes minimally to the calculated Pmax value. As shown in Figure 3D, the average fluorescence on times (⟨τon⟩) of α-LA molecules are significantly longer than the camera exposure time (30 ms). Thus, although undercounting cannot be entirely discounted as contributing to the disparity between single molecule and ensemble Pmax values, the more likely explanation is that P max values extracted from ensemble experimental

⟨τon⟩−1 =

1 ∞

∫0 τfon (τ ) dt

= kdes (7)

where fon(τ) is the probability density function of τon. Shown in Figure 3D are the ⟨τon⟩−1 values for α-LA at 2 nM, 5 nM, and 10 nM, respectively. It is clear that ⟨τon⟩−1 is independent of the αLA concentration as expected; however, their values are much greater than our fitted kdes. We speculate that kdes is small compared to ⟨τon⟩−1 because adsorption sites on nylon-6,6 films require time to recover by some slow but unknown mechanism other than protein photobleaching or denaturation. We first examined the rate of photobleaching of our fluorescently label Alexa 555. Control experiments (Figure S8 in the Supporting Information) showed that the rate of photobleaching was ∼0.1 s−1, which is slower than the measured ⟨τon⟩−1; and the change of laser power did not seem to affect the ⟨τon⟩−1 (Figure S9 in the Supporting Information). Therefore, photobleaching is only responsible for part of the fast disappearance of fluorescent signals. It is well-known that α-LA unfolds upon adsorption onto hydrophobic surfaces, such as the oil−water interface55 or polystyrene nanospheres.56 It is thus possible that α-LA also unfolds on the nylon-6,6 surface. However, the rate of unfolding is orders of magnitude slower than our ⟨τon⟩−1;57,58 and moreover, protein unfolding does not lead to the disappearance of fluorescent signal (Figure S10 in the Supporting Information). By ruling out photobleaching and unfolding, we speculate that the adsorbed α-LA molecules only reside on the nylon film for short period of time. The desorption of α-LA lead to the fast disappearance of fluorescent signal, i.e., large ⟨τon⟩−1. However, the adsorption site on the nylon film E

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 4. α-LA adsorption onto porous nylon-6,6 films. (A) SEM of porous nylon-6,6 film. Micrometer sized pores are shown as black domains in this image. (B) AFM of porous nylon-6,6 film. Top plot is the line profile across one pore, which is marked by the red line. (C) Dependence of adsorption rate versus time. The α-LA concentrations were 2 nM (blue), 5 nM (green), and 10 nM (red), respectively. Each data point represents the average rate of adsorption within a 30 s observation window; solid lines are global fits using eq 5, and error bars are standard deviation. (D) ⟨τon⟩−1 under various α-LA concentrations. Each data bar is the inverse averaged on-time of all protein molecules under the same conditions. The error bars represent the standard error of the mean.

(1.8 ± 0.1) × 10−4 s−1, and (3 ± 1) × 10−4 molecule nm−2, respectively (Table 1). Here the Pmax was normalized by the projected surface area of nylon-6,6 film. Compared to the flat nylon films, the adsorption rate constant (kads) decreased by a factor of 4, the desorption rate constant (kdes) was roughly the same, the equilibrium constant (Keq) decreased by a factor of 4, and the Gibbs free energy (ΔGads) for adsorption increased to −1.3 ± 0.8 kJ mol−1. The decreased adsorption constant may stem from the increased steric repulsion of curved surfaces in pores. However, such steric repulsion did not seem to affect the desorption of adsorbed protein molecules since the desorption rate constants for the porous and flat nylon-6,6 films were similar. The total number of adsorption sites (Pmax) increased by a factor of 3 due to the larger surface area on porous nylon films as expected. The increase of Gibbs free energy by a factor of 3.6 suggested that the adsorption of protein onto porous films was energetically less favorable. Once again, the ⟨τon⟩−1 for porous film (Figure 4D) is independent to the α-LA concentration in bulk solution, and the absolute values were close to those from flat nylon-6,6 films (Figure 3D).

required a much longer time to recover before it was available for another adsorption event, leading to the small kdes in our fitting. Thus, it remains a task for future simulations or experiments to fully understand the desorption process and its effect on the nylon film. α-LA Adsorption Kinetics on Porous Nylon-6,6 Films. In contrast to flat films, porous nylon-6,6 films contained more sites for α-LA adsorption, but adsorption was energetically less favorable (Figure 4). Porous nylon-6,6 films were synthesized by spin coating nylon-6,6 in 9:1 v/v formic acid/ethanol mixture. The pores were generated due to the phase separation in the binary solvent.59,60 To eliminate the potential influence of the solvent residue in the spin coating process, we intensively rinsed the films with HEPES buffer and then water. Even though there still might be traces of solvent left during the measurements, their effects were negligible. The SEM image (Figure 4A) shows that the generated pores are a few micrometers to tens of micrometers in size, and some of the pores are adjacent to each other, forming even larger cavities. The average depth of pores ranged from 15 to 20 nm (Figure 4B), smaller than the thickness of film. Our single-molecule measurements were not influenced by the existence of the porous structures, because nylon-6,6 is highly hydrophilic, and the aqueous solutions can wet the entire pore. Therefore, TIRF is still achieved at the nylon−solution interface even inside pores. The intensities as well as the signal-to-noise ratios for the adsorption events on porous nylon films were similar to that of the flat films. Compared to the flat nylon-6,6 films, the initial rate of adsorption on porous films decreased only slightly (Figure 4C). By globally fitting the time dependent rate of adsorption with eq 5, kads, kdes, Pmax were determined to be (3 ± 1) × 10−4 nM−1 s−1,



CONCLUSION We used a single-molecule fluorescence imaging approach to study protein−polymer film interactions. We verified that the αLA molecules only interact at the surface of nylon-6,6 films and they adsorb to nonspecific surface sites. By counting the number of newly adsorbed α-LA molecules, we were able to directly quantify the rate of α-LA adsorption onto both flat and porous nylon films. The rate of adsorption exhibited clear dependences on the α-LA concentration and time, and the adsorption processes took more than 1 h to reach equilibrium. In addition, F

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(5) Svec, F.; Fréchet, J. M. J. Science 1996, 273, 205−211. (6) Lvov, Y.; Ariga, K.; Ichinose, I.; Kunitake, T. J. Am. Chem. Soc. 1995, 117, 6117−6123. (7) Rabe, M.; Verdes, D.; Seeger, S. Adv. Colloid Interface Sci. 2011, 162, 87−106. (8) Kim, J.-H.; Yoon, J.-Y. Encyclopedia of Surface and Colloid Science; Marcel Dekker: New York, 2002; pp 4373−4381. (9) Zhang, Z.-H.; Feng, C.-L. Biotechnol. J. 2007, 2, 743−751. (10) Noinville, V.; Vidal-Madjar, C.; Sebille, B. J. Phys. Chem. 1995, 99, 1516−1522. (11) Liu, S. X.; Kim, J.-T. JALA 2009, 14, 213−220. (12) Stenberg, E.; Persson, B.; Roos, H.; Urbaniczky, C. J. Colloid Interface Sci. 1991, 143, 513−526. (13) Benesch, J.; Hungerford, G.; Suhling, K.; Tregidgo, C.; Mano, J. F.; Reis, R. L. J. Colloid Interface Sci. 2007, 312, 193−200. (14) Hirn, R.; Schuster, B.; Sleytr, U. B.; Bayerl, T. M. Biophys. J. 1999, 77, 2066−2074. (15) Anikin, K.; Röcker, C.; Wittemann, A.; Wiedenmann, J.; Ballauff, M.; Nienhaus, G. U. J. Phys. Chem. B 2005, 109, 5418−5420. (16) Kwok, K. C.; Yeung, K. M.; Cheung, N. H. Langmuir 2007, 23, 1948−1952. (17) Kisley, L.; Chen, J.; Mansur, A. P.; Shuang, B.; Kourentzi, K.; Poongavanam, M.-V.; Chen, W.-H.; Dhamane, S.; Willson, R. C.; Landes, C. F. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 2075−2080. (18) Tauzin, L. J.; Shuang, B.; Kisley, L.; Mansur, A. P.; Chen, J.; de Leon, A.; Advincula, R. C.; Landes, C. F. Langmuir 2014, 30, 8391− 8399. (19) McLoughlin, S. Y.; Kastantin, M.; Schwartz, D. K.; Kaar, J. L. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 19396−19401. (20) Zareh, S. K.; Wang, Y. M. Microsc. Res. Tech. 2011, 74, 682−687. (21) Huang, B.; Wu, H.; Kim, S.; Zare, R. N. Lab Chip 2005, 5, 1005− 1007. (22) Lu, H. P.; Xun, L.; Xie, X. S. Science 1998, 282, 1877−1882. (23) Daniels, C. R.; Reznik, C.; Landes, C. F. Langmuir 2010, 26, 4807−4812. (24) Reznik, C.; Estillore, N.; Advincula, R. C.; Landes, C. F. J. Phys. Chem. B 2009, 113, 14611−14618. (25) Gasser-Ramirez, J. L.; Harris, J. M. Anal. Chem. 2010, 82, 5743− 5750. (26) Michalet, X.; Weiss, S.; Jäger, M. Chem. Rev. 2006, 106, 1785− 1813. (27) Xu, W.; Kong, J. S.; Yeh, Y.-T. E.; Chen, P. Nat. Mater. 2008, 7, 992−996. (28) Roeffaers, M. B. J.; Sels, B. F.; Uji-i, H.; De Schryver, F. C.; Jacobs, P. A.; De Vos, D. E.; Hofkens, J. Nature 2006, 439, 572−575. (29) Shen, H.; Zhou, X.; Zou, N.; Chen, P. J. Phys. Chem. C 2014, 118, 26902−26911. (30) Kastantin, M.; Langdon, B. B.; Chang, E. L.; Schwartz, D. K. J. Am. Chem. Soc. 2011, 133, 4975−4983. (31) Kastantin, M.; Keller, T. F.; Jandt, K. D.; Schwartz, D. K. Adv. Funct. Mater. 2012, 22, 2617−2623. (32) Langdon, B. B.; Mirhossaini, R. B.; Mabry, J. N.; Sriram, I.; Lajmi, A.; Zhang, Y.; Rojas, O. J.; Schwartz, D. K. ACS Appl. Mater. Interfaces 2015, 7, 3607−3617. (33) McKinnon, B. T.; Avis, K. Am. J. Health Syst. Pharm. 1993, 50, 1921−1936. (34) Ho, W.; Sirkar, K. Membrane Handbook; Van Nostrand Reinhold: New York, 1992. (35) Jan, D. E.; Raghavan, S. Colloids Surf., A 1994, 92, 1−7. (36) Zaluzec, E. J.; Gage, D. A.; Allison, J.; Watson, J. T. J. Am. Soc. Mass Spectrom. 1994, 5, 230−237. (37) Skordoulis, C. D.; Makropoulou, M.; Serafetinides, A. A. Appl. Surf. Sci. 1995, 86, 239−244. (38) Shuang, B.; Chen, J.; Kisley, L.; Landes, C. F. Phys. Chem. Chem. Phys. 2014, 16, 624−634. (39) Parthasarathy, R. Nat. Methods 2012, 9, 724−726. (40) Betzig, E.; Patterson, G. H.; Sougrat, R.; Lindwasser, O. W.; Olenych, S.; Bonifacino, J. S.; Davidson, M. W.; Lippincott-Schwartz, J.; Hess, H. F. Science 2006, 313, 1642−1645.

we proposed a simple monolayer adsorption model that is composed of a pseudo-second-order adsorption and a pseudofirst-order desorption processes, from which some important kinetic and thermodynamic parameters for the adsorption were obtained. By creating micrometer sized pores on nylon-6,6 films, the adsorption rate constants decreased by a factor of 4, although the total number of adsorption sites increased by a factor of 3. The Gibbs free energy for α-LA adsorption onto flat and porous nylon-6,6 films was measured to be −4.7 ± 0.9 kJ mol−1 and −1.3 ± 0.8 kJ mol−1, respectively, suggesting rougher surfaces were energetically less favorable for the adsorption. The current work reveals that there are still many questions remaining about the adsorption of proteins on polymers. For example, there is an apparent disparity between ensemble and single molecule adsorption loading capacity, one explanation for which is the existence of higher order adsorption kinetics at high protein concentrations. Exploring higher order interaction mechanisms is an exciting question for future work that combines more complicated single molecule measurement conditions, better photon detectors, and more complex models for multiscalar adsorption. Nevertheless, it is clear that the single molecule fluorescence approach is a powerful tool to understand the mechanisms of protein−polymer interactions.



ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b04081. Detailed procedures of the film preparation and characterization, procedures for protein labeling, evanescence field of TIRF microscopy, additional results on the rhodamine B molecules sandwiched within nylon-6,6 films, shapes of the point spread functions for the adsorbed α-LAs, singlemolecule lysozyme adsorption onto flat nylon-6,6 films, procedure and residual plots for the global fits, power dependent on-time, and the influence of protein unfolding (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: cfl[email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research was supported by 3M, the NSF (Grant CHE1151647), and the Welch Foundation (Grant C-1787). L.K. acknowledges the NSF for Graduate Research Fellowship 0940902. We thank the Willson research group at University of Houston for providing the protein samples and the Link group at Rice University for discussions.



REFERENCES

(1) Chen, H.; Yuan, L.; Song, W.; Wu, Z.; Li, D. Prog. Polym. Sci. 2008, 33, 1059−1087. (2) Wang, R.; Chen, W.; Meng, F.; Cheng, R.; Deng, C.; Feijen, J.; Zhong, Z. Macromolecules 2011, 44, 6009−6016. (3) Banerjee, I.; Pangule, R. C.; Kane, R. S. Adv. Mater. 2011, 23, 690− 718. (4) Schwarz, J. A., Contescu, C. I., Eds. Surfaces of Nanoparticles and Porous Materials; Marcel Dekker: New York, 1999. G

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry (41) Rust, M.; Bates, M.; Zhuang, X. Nat. Methods 2006, 3, 793−795. (42) Bates, M.; Huang, B.; Dempsey, G. T.; Zhuang, X. Science 2007, 317, 1749−1753. (43) Xu, W.; Shen, H.; Kim, Y. J.; Zhou, X.; Liu, G.; Park, J.; Chen, P. Nano Lett. 2009, 9, 3968−3973. (44) Chen, J.; Bremauntz, A.; Kisley, L.; Shuang, B.; Landes, C. F. ACS Appl. Mater. Interfaces 2013, 5, 9338−9343. (45) Rust, M. J.; Bates, M.; Zhuang, X. Nat. Methods 2006, 3, 793−796. (46) Huang, B.; Wang, W.; Bates, M.; Zhuang, X. Science 2008, 319, 810−813. (47) Satterfield, C. N. Heterogeneous Catalysis in Practice; McGraw-Hill: New York, 1980. (48) Rodriguez-Emmenegger, C.; Kylián, O.; Houska, M.; Brynda, E.; Artemenko, A.; Kousal, J.; Alles, A. B.; Biederman, H. Biomacromolecules 2011, 12, 1058−1066. (49) Absolom, D. R.; Zingg, W.; Neumann, A. W. J. Biomed. Mater. Res. 1987, 21, 161−171. (50) Beeskow, T.; Kroner, K. H.; Anspach, F. B. J. Colloid Interface Sci. 1997, 196, 278−291. (51) Wei, Q.; Becherer, T.; Angioletti-Uberti, S.; Dzubiella, J.; Wischke, C.; Neffe, A. T.; Lendlein, A.; Ballauff, M.; Haag, R. Angew. Chem., Int. Ed. 2014, 53, 8004−8031. (52) Vogler, E. A. Adv. Colloid Interface Sci. 1998, 74, 69−117. (53) Xu, W.; Kong, J. S.; Chen, P. J. Phys. Chem. C 2009, 113, 2393− 2404. (54) Kou, S. C.; Cherayil, B. J.; Min, W.; English, B. P.; Xie, X. S. J. Phys. Chem. B 2005, 109, 19068−19081. (55) Zhai, J.; Hoffmann, S. V.; Day, L.; Lee, T.-H.; Augustin, M. A.; Aguilar, M.-I.; Wooster, T. J. Langmuir 2012, 28, 2357−2367. (56) Engel, M. F. M.; van Mierlo, C. P. M.; Visser, A. J. W. G. J. Biol. Chem. 2002, 277, 10922−10930. (57) Wehbi, Z.; Pérez, M.-D.; Dalgalarrondo, M.; Sánchez, L.; Calvo, M.; Chobert, J.-M.; Haertlé, T. Mol. Nutr. Food Res. 2006, 50, 34−43. (58) Kita, N.; Kuwajima, K.; Nitta, K.; Sugai, S. Biochim. Biophys. Acta, Protein Struct. 1976, 427, 350−358. (59) Wijmans, J. G.; Baaij, J. P. B.; Smolders, C. A. J. Membr. Sci. 1983, 14, 263−274. (60) van de Witte, P.; Dijkstra, P. J.; van den Berg, J. W. A.; Feijen, J. J. Membr. Sci. 1996, 117, 1−31.



NOTE ADDED AFTER ASAP PUBLICATION This paper was published ASAP on September 29, 2016 with an incorrect rate constant in the second full paragraph on page E and in Table 1 due to production error. The corrected version reposted to the Web on September 30, 2016.

H

DOI: 10.1021/acs.analchem.5b04081 Anal. Chem. XXXX, XXX, XXX−XXX