Gold Binding Peptide Identified from Microfluidic Biopanning: An

Article Cite This: Langmuir XXXX, XXX, XXX−XXX

pubs.acs.org/Langmuir

Gold Binding Peptide Identified from Microfluidic Biopanning: An Experimental and Molecular Dynamics Study Dong Jae Lee,† Hyun Su Park,† Kunmo Koo,†,‡ Jeong Yong Lee,†,‡ Yoon Sung Nam,†,§ Wonhee Lee,*,§,∥,⊥ and Moon Young Yang*,§

Langmuir Downloaded from pubs.acs.org by SAN FRANCISCO STATE UNIV on 01/02/19. For personal use only.

†

Department of Materials Science and Engineering, §KAIST Institute for NanoCentury, ∥Graduate School of Nanoscience and Technology, and ⊥Department of Physics, Korea Advanced Institute of Science and Technology, 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea ‡ Center for Nanomaterials and Chemical Reactions, Institute for Basic Science (IBS), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea S Supporting Information *

ABSTRACT: Biopanning refers to the processes of screening peptides with a high affinity to a target material. Microfluidic biopanning has advantages compared to conventional biopanning which requires large amounts of the target material and involves inefficient multiple pipetting steps to remove nonspecific or low-affinity peptides. Here, we fabricate a microfluidic biopanning system to identify a new gold-binding peptide (GBP). A polydimethylsiloxane microfluidic device is fabricated and bonded to a glass slide with a gold pattern that is deposited by electron-beam evaporation. The microfluidic biopanning system can provide high adjustability in the washing step during the biopanning process because the liquid flow rate and the resulting shear stress can be precisely controlled. The surface plasmon resonance analysis shows that the binding affinity of the identified GBP is comparable to previously reported GBPs. Moreover, molecular dynamics simulations are performed to understand its binding affinity against the gold surface in detail. Theoretical calculations suggest that the association and dissociation rates of the GBPs depend on their sequence-dependent conformations and interactions with the gold surface. These findings provide insight into designing efficient biopanning tools and peptides with a high affinity for various target materials.

■

INTRODUCTION

highly affected by the mechanical force applied to the peptides.2−4 Although optimization of the washing condition is essential, there is no practical guideline to efficiently screen M13 phages expressing high-affinity peptides,5,6 which are selected by manual washing processes, such as pipetting and swirling the well plates. For the precise regulation of experimental conditions, microfluidic systems have been adopted to control the washing

Phage display is an effective means to identify the sequence of peptides that can specifically interact with target antigens, proteins, cells, and inorganic materials.1 M13 phage libraries displaying random peptides in pVIII or pIII coat proteins have been widely used for biopanning, which typically involves four major steps: target material immobilization, incubation of the phage library with the target material, removal of nonspecific and low-affinity phages from the target material, and recovery of the target-bound phages for the next round of selection. Among these four steps, the washing step has a critical impact on the resulting peptides because the affinity of peptide is © XXXX American Chemical Society

Received: July 28, 2018 Revised: December 18, 2018

A

DOI: 10.1021/acs.langmuir.8b02563 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 1. Fabrication and working principle of the microfluidic biopanning platform. (A) Fabrication process of a microfluidic biopanning platform. (B) Fabricated device (left) and schematic illustration of the biopanning process (right), where the blue arrows represent the strength of the flow rate. × 2.5 cm microscope slide. The substrate was then exposed to UV light through a patterned photomask using a mask aligner (MIDAS1600, Midas Systems, Daejeon, Republic of Korea), and the patterned photoresist layer was developed for 1 min. After the photoresist was patterned on the microscope slide, a chromium layer of 10 nm and a gold layer of 65 nm were deposited on the substrate using an electronbeam (E-beam) evaporator. After the photoresist layer was removed by a lift-off process with sonication in acetone for 5 min, a PDMS microfluidic channel was aligned to the gold pattern and plasmabonded (PDC-002, Harrick Plasma, Ithaca, NY, USA) to the substrate. The details are described in the Supporting Information and our previous paper.10 Microfluidic Biopanning. The biopanning process was performed using the gold-patterned microfluidic chip with NEB Ph.D.-7 phage display peptide library kit. Ten microliters of phage solution (2 × 1012 pfu mL−1) in 0.1 vol % PBST (PBS (pH 7.4) containing 0.1 vol % Tween 20) was loaded into the microfluidic biopanning platform at a flow rate of 1 μL min−1 at room temperature. Then PBST (0.1 vol % Tween 20 in the first round and 0.5 vol % Tween 20 from the second round) was introduced at 0.5 mL min−1 for 30 min to remove nonspecifically or weakly bound phages. The remaining high-affinity phages were eluted by injecting an elution buffer (0.2 M glycine-HCl (pH 2.2) with 1 mg mL−1 BSA) at 0.1 mL min−1 for 10 min, and the eluate was neutralized with 150 μL of 1 M Tris-HCl (pH 9.1). Five rounds of biopanning were performed, and the eluted phages from each round were amplified by incubation with ER2738 host cells. Finally, the phage peptide sequence was analyzed by sequencing the phage plasmid DNA. Surface Plasmon Resonance Analysis. SPR analysis was performed using a Biacore 3000 (GE Healthcare, Pittsburgh, PA, USA) at 25 °C for the three peptides. To measure the binding kinetics of GBPs to the gold surface, Sensor Chip Au (GE Healthcare, Pittsburgh, PA, USA) was used. The peptides were dissolved in PBS and filtered using a 0.22 μm syringe filter. The association, equilibrium, and dissociation phases were monitored by injecting GBPs into the SPR chip at a flow rate of 10 μL min−1. The regeneration of the chip surface was performed by injecting a 100 mM NaOH solution. The parameters of kinetics, such as association rate constant (kon), dissociation rate constant (koff), and dissociation constant (Kd), were determined by analyzing the obtained sensorgrams with the BIAevaluation software (BIAcore, version 4.1). Crystallographic Orientation of Gold Surfaces. To identify the crystalline structure of the gold surface, the crystallographic orientations were analyzed using a transmission electron microscope (TEM) (Titan ETEM G2, FEI). The TEM specimen were prepared using a focused ion beam (FIB) (Helios NanoLab DualBeam FIB System, FEI) technique. For the microfluidic biopanning device, the gold layer was passivated by amorphous carbon and platinum layer to protect the thin lamella of the gold layer prior to etching the substrate.

stringency during the phage selection.7−11 Microfluidic biopanning systems not only require a small sample volume but also can carry out multiple steps in parallel or series in a single device. Because of their advantages, microfluidic biopanning systems have received increasing attention for screening peptides that exhibit a high binding affinity against target materials. However, most microfluidic biopanning platforms have focused on targeting biological molecules, despite the increasing interest in high binding affinity peptides against inorganic materials, such as gold or iridium oxide.12−14 We report here the fabrication of a microfluidic biopanning platform for inorganic materials, mainly focusing on gold because gold-binding peptides (GBPs) have high potential for various applications in diagnostics, bioimaging, and drug delivery.15−18 Moreover, their binding mechanisms on the gold surface are well-investigated by pioneering works.19−25 The fabricated microfluidic system has a simple structure compared to conventional microfluidic biopanning systems that require additional parts, such as magnetic beads coated with the target material and a ferromagnetic layer for enhancing the local magnetic field and capturing the magnetic beads.7,8 By using the fabricated system with the optimal flow rate reported in our previous work,10 we obtained a new GBP sequence. The binding affinities of the identified GBP and two previously reported GBPs to the gold surface were analyzed by surface plasmon resonance (SPR) analysis. Moreover, molecular dynamics (MD) simulations were carried out to understand their distinctive properties in association and dissociation rates depending on the peptide sequences.

■

EXPERIMENTAL SECTION

Materials. Hexamethyldisilazane (HMDS) and AZ GXR 601 (positive photoresist) were purchased from AZ Electronic Materials (Luxembourg). Polydimethylsiloxane (PDMS) and SU-8 3025 (negative photoresist) were obtained from Dow Corning (Midland, MI, USA) and MicroChem Corp. (Westborough, MA, USA), respectively. Gold and chromium were obtained from Materion Corp. (Mayfield Heights, OH, USA), and Ph.D.-7 phage display peptide library was obtained from New England Biolabs (NEB, Ipswich, MA, USA). Bovine serum albumin (BSA), glycine, Tween 20, polyethylene glycol, and all organic solvents were purchased from Sigma-Aldrich (St. Louis, USA). The peptides for the SPR analysis were synthesized and purchased from Peptron (Daejeon, Republic of Korea). Fabrication of the Microfluidic Biopanning Platform. Positive photoresist with 1 μm thickness was spin-coated on a 2.5 B

DOI: 10.1021/acs.langmuir.8b02563 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir Next, gallium beam was used to etch the substrate, leaving behind a thin specimen supported by bulk materials on the two opposite sides. Afterward, the cross section of the substrate was observed using a TEM. Molecular Dynamics Simulations. All MD simulations of GBPs were carried out using the software package GROMACS (5.1.4)26 patched with the PLUMED (2.3.0) plugin.27 Replica exchange with solute tempering (REST) simulations,28−30 a Hamiltonian-based REMD method, were performed with two different systems for each peptide: the isolated peptide in solution and the peptide adsorbed on the aqueous gold [111] surface. CHARMM22*31,32 and the modified TIP3P33,34 force fields were used for the peptide and water, respectively, and the peptide−gold interaction was described by GolP-CHARMM.35 VMD program was used for the visualization.36 The details of the simulations and analyses are described in the Supporting Information.

■

RESULTS AND DISCUSSION Microfluidic Device Fabrication. The fabrication procedures of the microfluidic platform for gold biopanning are described in Figure 1A. The gold-patterned surface was fabricated via photolithography and E-beam deposition on a microscope glass slide. The gold layer (65 nm) was deposited after the chromium layer (10 nm) deposition. Relatively thick gold layer was deposited to reduce the effect of the chromium layer on the surface properties of the gold layer. The chromium layer was deposited as an adhesion layer because of the poor adhesion between gold and glass. In principle, the microfluidic biopanning platform is applicable to various polymeric and inorganic materials because the gold layer can be easily replaced with various thin film materials that can be deposited by conventional nanofabrication techniques such as spin coating, chemical vapor deposition (CVD), and physical vapor deposition (PVD). The PDMS channel was aligned to the gold pattern and bonded to the glass slide after air plasma activation. The channel length is approximately 33 mm and the cross-sectional dimension is 60 × 500 μm (height × width) (Figure 1B). The overall device size is 2.5 × 2.5 cm. The flow rate of fluids can be precisely controlled with a syringe pump. A flow rate of 1 μL min−1 for 30 min was used during the phage library incubation, which was increased to 0.5 mL min−1 for 30 min during the washing process followed by the elution process at 0.1 mL min−1 for 10 min. These optimized flow rates had been experimentally determined in our previous study.10 The cross-section image of the fabricated device shows the three layers: glass, chromium, and gold (Figure 2A). TEM and X-ray diffraction (XRD) analyses for the gold layer suggest that the gold [111] surface is likely to be exposed on the majority of the microfluidic device surface (Figure 2 and Figure S1), although the overall structure is polycrystalline because of the nature of the deposition method.37 Identification of a New Gold Binding Peptide. The biopanning process was carried out to identify a GBP using the fabricated microfluidic system. Figure 3A shows the number of eluted phages during the phage selection. In the first round, a washing buffer containing a lower concentration of surfactant (0.1 vol % PBST) was used, which allows a higher diversity in the eluted phage pool and preserves more potential highaffinity bacteriophages. The surfactant concentration was then increased to 0.5 vol % PBST from the second round so that only phages having high affinity to the target remain. The effect of an increase in surfactant concentration is represented by the drastic decrease in the number of eluted phages from the first

Figure 2. Structural analysis of the microfluidic biopanning platform. (A) TEM image of the cross section of the substrate (scale bar = 50 nm), where the inset figure shows the crystallographic orientation of gold layer at the surface. (B) Selective area electron diffraction (SAED) pattern of the thin film and (C) the fast Fourier transform image.

Figure 3. Biopanning results from the microfluidic gold biopanning platform. (A) Enrichment of phage particles binding to the gold surface (n = 3). (B) Peptide sequences from the pIII protein of the clones eluted after round 5.

to the second round (Figure 3A). As the rounds progress, the proportion of high-affinity bacteriophages increases and fewer bacteriophages are removed from the surface after the washing step. Thus, the number of eluted bacteriophages increases as the rounds continue. For each biopanning round, the sequences displayed on the pIII protein of the phages were investigated by randomly picking 12−25 plaques from the eluted phage pool. After the five rounds, 7 out of 12 clones exhibit the glycine-threonine-glycine-serine-glutamine-alanineserine (GTGSQAS; hereafter referred to as MAP, which stands for microfluidic-biopanning Au Peptide) sequence (Figure 3B). The result of DNA sequencing from round 2 to round 4 shows that there was no bias (see the Figure S2A, Supporting Information), although a limited number of plaques were selected. A control experiment was also performed using another microfluidic device with the same initial phage display library, which has no gold pattern on the surface, to exclude the possibility that the identified peptide was obtained by the interaction with PDMS, glass, or PEG coating. After the five C

DOI: 10.1021/acs.langmuir.8b02563 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

peptide can interact more quickly with the surface than the longer peptide,49,50 which would also result in fast kinetic rates. Figure S5 (Supporting Information) shows the saturation of the equilibrium constant against the peptide concentration, which is smaller than the maximum response.51 From the SPR analyses, the binding affinities (Kd) of all three peptides were obtained (Table 1): 2.83, 2.28, and 3.67

rounds of the biopanning process, 7 out of 12 clones exhibited the glycine-valine-glutamate-glycine-histidine-lysine-proline (GVEGHKP) sequence, which is entirely different from the sequence of the MAP (see Figure S2B, Supporting Information). In fact, the GVEGHKP sequence has been reported to have an affinity to Fe3O4.38 This sequence may have come from the interaction with the stainless steel syringe tip used to inject liquid into the microfluidic biopanning platform. These results indicate that MAP was screened because of its binding affinity to the gold surface, although further studies should be followed to demonstrate its specificity. Surface Plasmonic Resonance Analysis. To evaluate the binding affinity of the MAP against the gold surface, the SPR analysis was performed, where we confirmed that the [111] surface was the most dominant in the SPR chip, similar to that in the microfluidic device, by TEM and XRD analyses (see the Figures S3 and S4, Supporting Information). In addition to the MAP, two other previously reported GBPs with the same sequence length were also analyzed for comparison: VSGSSPDS (p8#9)14 and KHKHWHWG (Z1).39 The SPR sensorgrams for three peptides are shown in Figure 4. It shows somewhat fast kon and koff rates compared to

Table 1. Kinetic Parameters of Three GBPs by SPR Analysis kon [M−1 s−1] koff [s−1] Kd [μM]

MAP

p8#9

Z1

2.8 × 104 7.92 × 10−2 2.83

5.31 × 104 1.21 × 10−1 2.28

1.64 × 104 6.01 × 10−2 3.67

μM for the MAP, p8#9, and Z1, respectively. For Z1, the obtained binding affinity is in good agreement with that in the previous study, ΔG = −31.3 kJ/mol, where it was estimated by quartz crystal microbalance analysis.21,52 Despite the similar binding affinities, they exhibit different properties depending on their sequences in terms of kon and koff. The results of SPR analyses shows that p8#9 has the highest kon value, while Z1 has the lowest koff value, and MAP exhibits moderate kon and koff values (Table 1). To understand these different binding properties in detail, we carried out MD simulations. Structural Analysis of Peptides by MD Simulations. REST MD simulations for each peptide adsorbed on the gold surface and that in solution were performed to investigate the relationship between the sequence and its binding affinity against the gold surface, where the most frequently exposed [111] surface was used for the calculations. Figure 5A shows

Figure 5. Central structures of the largest cluster of each GBP adsorbed on the gold [111] surface (A−C) and in solution (D−F), where the size of side chains corresponds to their contact ratio with the gold surface in (A−C). The water molecules are not shown for clarity.

the central structure of the largest cluster obtained by cluster analysis for each GBP adsorbed on the gold surface. For each peptide, the fraction of contact residues with the gold surface was calculated to estimate the interactions of each residue on the binding affinity (Table 2), which should be related to the koff value. For all three peptides, the number of their contact residues is similar: (1) four high-contact residues and four medium-contact residues for MAP, (2) three high-contact residues, four medium-contact residues, and one low-contact residue for p8#9, and (3) four high-contact residues, three medium-contact residues, and one low-contact residue for Z1, respectively. Despite similar contact populations, the koff values can vary depending on sequences because each amino acid has a different binding affinity to the solid surface depending on their side chains. For the gold [111] surface, it has been

Figure 4. SPR sensorgrams of (A) MAP, (B) p8#9, and (C) Z1, respectively.

previous GBP studies,40−42 although fast kinetic rates have been reported for other systems.43−47 These fast kinetic rates may result from the untreated gold surface and/or the short peptide length. Compared to the bare gold, other types of gold substrate, for example, gold nanoparticles,40 not only possess different surface properties but also inevitably contain ligands on its surface during the synthesis process,48 which could affect the diffusion and binding of the GBPs. Moreover, the smaller D

DOI: 10.1021/acs.langmuir.8b02563 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

simulations explain the different tendency of koff and kon values depending on the sequence among GBPs. In comparison, Walsh and co-workers have reported different propensity for the case of other GBP, AuBP1, on three different gold facet;,24 that is, the conformations of AuBP1 were not significantly changed between on the gold surface and in solution, and the contact ratio of each residue varied depending on the facets. This inconsistency may result from the different peptide sequences because the binding property of a peptide is determined by its sequence: the intrinsic binding propensity of each residue and the conformation as a whole. The different facets also should affect the behavior of the peptide. Although it is difficult to make a conclusion because of the limited sample number, our results suggest that the conformational change of the peptide at the interface can be one of many factors that determine the binding affinity, particularly on the kon value. Therefore, the conformation of the peptide depending on the surrounding environment should be considered for the rational design of targeting peptides.

Table 2. Calculated Contact Ratio (%) for Each Residue from MD Simulations, Where the Criteria for High-, Medium-, and Low-Contact Residues Are >75%, >50%, and