Article pubs.acs.org/Langmuir
Fibrillar Structures Formed by Covalently Bound, Short, β‑Stranded Peptides on Self-Assembled Monolayers Jason W. Dugger and Lauren J. Webb* Department of Chemistry, Center for Nano- and Molecular Science and Technology, and Institute for Cell and Molecular Biology, The University of Texas at Austin, 1 University Station, A5300, Austin, Texas 78712, United States S Supporting Information *
ABSTRACT: The ability to maintain or reproduce biomolecular structures on inorganic substrates has the potential to impact diverse fields such as sensing and molecular electronics, as well as the study of biological self-assembly and structure−function relationships. Because the structure and self-assembly of biomolecules are exquisitely sensitive to their local chemical and electrostatic environment, the goal of reproducing or mimicking biological function in an abiological environment, including at a surface, is challenging. However, simple and well-characterized chemical modifications of prepared surfaces can be used to tune surface chemistry, structure, electrostatics, and reactivity of inorganic materials to facilitate biofunctionalization and function. Here, we describe the covalent attachment of 13-residue βstranded peptides containing alkyne groups to a flat gold surface functionalized with an azide-terminated self-assembled monolayer through a Huisgen cycloaddition, or “click”, reaction. The chemical composition and structural morphology of these surfaces were characterized using X-ray photoelectron spectroscopy, grazing incidence angle reflection−absorption infrared spectroscopy, surface circular dichroism, and atomic force microscopy. The surface-bound β-strands self-assemble into antiparallel β-sheets to form fibrillar structures 24.9 ± 1.6 nm in diameter and 2.83 ± 0.74 nm in height on the reactive surface. The results herein provide a platform for studying and controlling the self-assembly process of biomolecules into larger supermolecular structures while allowing tunable control through chemical functionalization of the surface. Interest in the mechanisms of formation of fibrillar structures has most commonly been associated with neurodegenerative diseases, such as Alzheimer’s and Parkinson’s, but fibrils may actually represent the thermodynamic low-energy conformation of a much larger class of peptides and proteins. The protocol developed here is an important step toward uncovering not only the factors that dictate self-assembly but also the mechanisms by which this fibrillar class of superstructures forms.
■
finding surface chemistries that integrate the biomolecule with the surface in such a way that the chemical and electrostatic properties of the surface act in concert with the biomolecule to generate a novel biological/abiological material. To this end, our group has explored the use of self-assembled monolayers (SAM) on gold surfaces to facilitate the covalent attachment of structured α-helical peptides orientated parallel to the substrate.11 This method exploits the known structure and regularity of the SAM to induce a desired secondary structure in the biomolecules while preserving desirable properties characteristic of the metal surface such as conductivity.12,13 Here, we report an adaptation of this methodology to covalently bound, short (13-residue), β-stranded peptides to the gold surface, which in turn nucleate the self-assembly of large supermolecular fibrillar structures in a controlled way.
INTRODUCTION The chemical functionalization of inorganic surfaces and substrates with folded and functional biological molecules has the potential for a large variety of applications, including sensors, electrochemistry, and molecular electronics, as well as providing a template for studying and controlling the selfassembly of biomolecules into supermolecular structures. The immobilization of biomolecules on surfaces enables the formation and examination of biomolecular structures that would not be possible in solution, such as single-molecule studies,1 the formation of asymmetric and two-dimensional materials,2−4 and the generation of novel materials that incorporate both biological and abiological functions.5−7 A chief obstacle to the incorporation of biomolecules onto inorganic substrates is the preservation of the structure of the molecule, which is inherently related to its function and stability. One strategy that has been employed successfully is to cover the underlying inorganic substrate with a thick polymer, such as polyethylene glycol or polylysine, reducing the surface to a passive support.8−10 Our laboratory has been interested in © 2015 American Chemical Society
Received: December 23, 2014 Revised: February 20, 2015 Published: March 4, 2015 3441
DOI: 10.1021/la5049369 Langmuir 2015, 31, 3441−3450
Article
Langmuir
well as self-assembly into antiparallel β-sheets. Additionally, structural morphology and heterogeneity observed using atomic force microscopy (AFM) show that these β-sheets organize into larger fibrillar structures 24.9 ± 1.6 nm in diameter and 2.83 ± 0.74 nm in height. We hypothesize that single peptide strands covalently bound to the SAM act as nucleation sites for the formation of β-sheets, which then proceed to organize into fibrils in as little as 15 min. Samples with longer reaction times (18 h) contain similarly sized structures with a higher degree of surface coverage, creating a meshlike network of fibrils. These studies establish the capability to nucleate noncovalently assembled superstructures through the controlled immobilization of short peptides to inorganic substrates. This strategy inherently enables tuning of chemical reactivity, environmental conditions, and interface heterogeneity and generalization to other biomolecules.
Biological and synthetic fibrils and fibrillar structures have shown promise in applications such as sensors, ordered nanomaterials, mimics of extracellular matrices, and scaffolds for cell cultures.14−18 Biological fibrils are also of immense current medical interest because they are classically associated with neurodegenerative diseases such as Alzheimer’s, Parkinson’s, and type II diabetes.19−21 Because of this, there has been extensive investigation into the circumstances under which naturally occurring biological peptides spontaneously selfassemble into aggregates, fibrils, and plaques in certain tissues throughout the body, and it is widely believed that one or more aspects of this self-assembly process may be causing the disease state associated with that fibril. However, it has recently become clear that the formation of these fibrillar structures is not a unique property of a particular biological peptide, but instead seems to be a common feature of amide-based polypeptides in general.22 Indeed, recent computational and experimental work on the thermodynamic stability of fibrils and kinetic barriers to the assembly of these structures has suggested that fibrillar conformations may be the low-energy structure of polypeptides in general, regardless of sequence or length.23−26 These observations mean that the basic understanding of biological fibril formation, stability, and function is a research area that lies at a crossroads between materials science, biophysical chemistry, and medicine, influencing fields as diverse as human health and molecular electronics. Various factors that regulate fibril formation and growth, including kinetics, inhibition, promotion, sequence variation, and environmental conditions, have been investigated,27−31 and it has been observed repeatedly that there are a large variety of conformations that a single protein sequence can adopt.19 It has been shown that fibrils formed from amyloidogenic peptides of different sequences can adopt structures comprised of parallel or antiparallel β-sheets, and the resulting fibrils can be polymorphic.21,32−34 For instance, fibrils have been seen to adopt a triangular structure with a large central cavity,35 a thin ribbon-like structure with diameters ranging from 3.5 to tens of nanometers,36 and even nanotube-like structures over 100 nm in diameter resulting from the helical twisting of ribbon-like fibrils.30,37 Despite the extensive characterization of a variety of these fibrillar structures, there is significant ambiguity concerning how the complex interactions dictated by peptide sequence, hydrophobicity, hydrophilicity, and environment regulate the nucleation and self-assembly of small biomolecules into more complex supermolecular structures. Understanding these factors is a necessary prerequisite to generate controlled materials containing desired properties that can be exploited for interesting or novel purposes. Taking a surface-mediated approach to immobilizing these biomolecules allows for unprecedented control over a number of variables that can be used to possibly tune fibril growth, size, structure, surface density, and chemical reactivity. In the work described here, we demonstrate the self-assembly of superstructures using a bottom-up method to fibril formation through the covalent attachment of short (4.4 nm, 13-residue long) β-stranded peptides to a SAM on a gold surface. This is accomplished with a triazole linkage formed from a Huisgen cycloaddition “click” reaction between one or two alkyne groups on the peptide and adjacent terminal azide groups on the SAM. Grazing incidence angle reflection−absorption infrared spectroscopy (GRAS-IR), X-ray photoelectron spectroscopy (XPS), and surface circular dichroic (CD) spectroscopy confirm the presence of the peptide on reacted surfaces as
■
MATERIALS AND METHODS
Surface Preparation. Peptide-terminated SAM surfaces were prepared on silicon (111) wafers for GRAS-IR and XPS measurements, on quartz slides for surface CD experiments, and on gold-onmica substrates (150 nm of gold on mica, SPI Supplies) for AFM measurements. The different substrates were used to exploit their unique properties in each experiment, with silicon wafers used for their macroscopic flatness in the spectroscopic reflection experiments, quartz for its UV transparency in CD studies, and mica for its microscopic flatness in AFM. Silicon wafers (NOVA Electronic Materials) were 500 μm thick, polished on one side, sealed in N2(g), and stored in a dry N2(g) atmosphere glovebox until use. A 10 nm layer of chromium, followed by 100 nm of gold (99.95% pure), was vapor-deposited on the silicon wafers using a thermal evaporator-II (Denton) instrument at a pressure of 10−5 Torr. The samples were covered in clean-room-rated silicon wafer tape (ICROS TAPE) and cut into approximately 1 cm2 pieces with a programmable Disco 321 wafer-dicing saw. After removing the tape, the samples were rinsed in methanol, sonicated in acetone for 10 min, rinsed in high purity water (HPW) with an impedance of >18 MΩ cm (Barnstead NANOpure Diamond Life Science UV/UF) and then rinsed again in methanol, dried under a stream of N2(g), and stored in Parafilm-covered containers until use. This cleaning was done to remove any residue that may have been left behind by the wafer tape used during the cutting process. Quartz substrates precut into 2.5 × 0.95 cm slides were rinsed in ethanol and dried under N2(g) before the deposition of 3 nm of chromium, followed by 11 nm of gold, and subsequently stored in Parafilm-covered containers until use. The quartz samples had the thinnest layer of gold of all the substrates in order to increase transmission in the CD experiments. Previous studies have shown that SAMs are still formed on gold layers as thin as 7 nm,38 and SAMs are commonly used to functionalize gold surfaces on a variety of supporting substrates, including those employed in these studies. The choice of substrate and thickness of the gold layer is not expected to affect SAM formation or subsequent reaction with the peptides. All chemicals were purchased from Sigma-Aldrich unless otherwise noted. Gold substrates deposited on silicon were cleaned using a piranha solution (1:3 30% hydrogen peroxide/concentrated sulfuric acid) (Caution! Explosive in the presence of organic contaminants.) for 1 min; rinsed in a sequence of HPW, concentrated hydrochloric acid, HPW, and finally ethanol; and dried under a stream of N2(g). These surfaces were then annealed for 5 min using a H2(g) flame. Gold-onmica substrates were kept in manufacturer’s packaging (glass vials filled with argon gas, capped, and sealed with Parafilm) until use, where they were then annealed for 5 min with a H2(g) flame. Quartz substrates were only rinsed in ethanol and dried under N2(g) because the thinner layer of gold could not withstand the harsher cleaning methods described above. All subsequent steps were identical for the silicon-, quartz-, and mica-supported substrates. To prepare the SAM, the samples were submerged in a solution of 1 mM 11-azido-1undecanethiol diluted in anhydrous ethanol for 24 h in the dark to 3442
DOI: 10.1021/la5049369 Langmuir 2015, 31, 3441−3450
Article
Langmuir
qualitatively unchanged (254 scans). For the β6-reacted surfaces, a total of 410 scans were accumulated, and beyond background subtraction, no other data manipulation was done to the spectra. X-ray Photoelectron Spectroscopy. To further characterize the chemical composition of the surfaces, X-ray photoelectron spectra were collected with a Kratos Azis Ultra XPS equipped with a monochromatic Al Kα source illuminating the sample at 1486.5 eV. Samples were grounded using copper tape and introduced into the chamber at a pressure below 2 × 10−9 Torr. High-resolution spectra of the carbon 1s region were collected at a resolution of 0.1 eV with a 20 eV pass filter. Ejected photoelectrons were collected with a hemispherical electron energy analyzer positioned normal to the sample surface. Due to sample charging from photoelectron loss, the adventitious carbon peaks were detected at a higher than normal binding energy. To correct for this, the spectra were shifted to reposition the adventitious carbon peak to 284.5 eV. Spectral shifts and decompositions were done using a linear baseline followed by peak fitting to Gaussian curves using the CasaXPS software package. Atomic Force Microscopy (AFM). An Asylum Research MFP-3D AFM was used to image the dry, reacted, gold-on-mica substrates in ambient air. AFM cantilevers (Mikromasch) with typical probe radii of 8 nm, 65 kHz resonance frequencies, and 0.5 N/m force constants were used in tapping mode to minimize shear forces and tip−sample interactions. Height and width measurements were obtained by averaging three line profiles oriented normal to the long axis of each fibril. A total of 25 measurements for each reaction time point were taken from images 1 × 1 μm in size. All image processing and line profiles were done using the Gwyddion SPM software package.
fabricate a gold surface covered in 100% azide-terminated SAM. The 11 methylene groups of the thiol participate in van der Waals interactions with the hydrophobic portions of surrounding thiols, inducing an ordered SAM that leaves the azide groups exposed at the SAM/aqueous interface. Unbound thiols were removed by sequentially rinsing with HPW, dimethylformamide (DMF), HPW, and ethanol and dried under a stream of N2(g). Azide-terminated SAM surfaces were exposed to the peptide KLKXKLLLKXKLK (hereafter “β6”; WuXi AppTec). It has been shown that sequences of alternating leucine and lysine residues adopt β-stranded conformations in aqueous solutions and at air/water interfaces.39 The β6 peptide was designed to approximate this periodicity with the exception of the three sequential hydrophobic residues to minimize β-sheet formation in solution. Two unnatural propargylglycine residues (X; Wuxi AppTec) contained reactive alkyne groups for a Huisgen cycloaddition (click) reaction, where the alkyne groups in the peptide react with the azide groups at the surface to form triazole linkages using Cu(I) as a catalyst. Sodium ascorbate was used to reduce Cu(II) to Cu(I), and TBTA {tris[(1-benzyl-1H-1,2,3-triazol4-yl)methyl]amine} was present to stabilize the Cu(I) in the aqueous solution environment.40 Solutions with a total volume of 5 mL were prepared by adding 2:1 tert-butanol/HPW solvent, followed by TBTA (0.5 μmol), β6 peptide (0.5 μmol), sodium ascorbate (0.6 μmol), the azide-terminated surface, and finally CuSO4 (0.1 μmol). Before addition to the reaction solution, the peptide was vortexed and then stirred for 20 min in the 2:1 tert-butanol/HPW solvent at concentrations below 1.3 mg/mL to ensure complete solubility. After addition of all reagents, vials were immediately placed in an oven at 70 °C for reaction times ranging from 15 min to 18 h. Upon removal, samples were allowed to cool in the reaction solution for 10 min, and then any physisorbed reactants were removed by sequentially rinsing with 2:1 t-butanol/HPW, HPW, phosphate-buffered saline (PBS), HPW, and ethanol and then dried under a stream of N2(g). Dried samples were stored in the dark until measured. All control samples were azide-terminated surfaces that where immersed in the reaction solution lacking CuSO4 and its reducing agent, sodium ascorbate, to eliminate the presence of any Cu(I) catalyst. Grazing Incidence Angle Reflection−Absorption Infrared Spectroscopy. Surface vibrational spectroscopy was collected with a Bruker Vertex 70 FTIR spectrometer equipped with a A518/Q horizontal reflection unit (Bruker) for illuminating the sample at a grazing angle of 80° with respect to the surface normal. The sample chamber was purged with N2(g) for 1 h after samples were introduced to the instrument to reduce background noise from H2O and CO2. A total of 400 scans of p-polarized light were collected for each sample before and after reaction with the β6 peptide. A mercury cadmium telluride (MCT) detector collected signals from the 400-4000 cm−1 region, while an indium antimonide (InSb) detector collected signals from 1870 to 4000 cm−1 to take advantage of its higher sensitivity to the azide, methyl, and methylene stretches in that region. A bare gold substrate (cleaned and annealed as described above with each set of samples) was used as a common reference in order to obtain absolute differences in absorbance between samples. Following background subtraction, the spectra were flattened using a rubber band correction baseline function in the instrument’s OPUS software. Circular Dichroic Spectroscopy. The conformation of the β6 peptide in solution and on the surface was characterized using a JASCO J-815 CD spectrometer.12,41,42 Samples were illuminated between 190 and 250 nm at a scan rate of 50 nm/min, 1 nm resolution, and a 4 s response time. For solution studies, a 130 μM solution of peptide dissolved in HPW in a 1 mm quartz cuvette was used to collect three scans, using a HPW solution for background subtraction. Surface CD spectra were obtained by stacking three β6reacted quartz slides in a 1 cm cuvette filled with HPW, oriented normal to and facing the UV source. Unreacted azide-terminated quartz slides set up in the same manner were used for background subtraction. Due to poor UV transmission through the gold and relatively low surface concentration of peptide compared to solution, a large number of scans were acquired to establish the spectra. Scans of the background slides were taken until the spectra appeared to be
■
RESULTS Chemical Composition of Peptide-Reacted Surfaces. GRAS-IR spectra were acquired for azide-terminated surfaces before and after reaction with the β6 peptide and for control samples. Representative spectra of the C−H stretching region for the surfaces discussed here are shown in Figure 1. Azideterminated surfaces (Figure 1, green) were dominated by features at 2851 and 2924 cm−1, corresponding to symmetric and asymmetric methylene stretches, respectively.11,43,44 Peptide-terminated surfaces reacted for 30 min (Figure 1, blue) showed an asymmetric methyl stretch at 2958 cm−1, as well as an increase in intensity of the methylene stretches due to the presence of additional C−H groups contributed by the surface-bound peptides. The peptide-terminated surface reacted for 18 h (Figure 1, red) also showed an increase in methylene peak intensities along with the development of symmetric and asymmetric methyl peaks at 2870 and 2958 cm−1. The reacted surfaces were significantly different from control surfaces prepared by immersing the azide-terminated SAM surface in the Huisgen cycloaddition reaction solution without any Cu(I) catalyst (Figure 1, black). These surfaces displayed methylene stretches at 2857 and 2928 cm−1, indicating preservation of the SAM, but except for a small methyl peak at 2962 cm−1, corresponding to a small amount of physisorbed peptide, they did not show any modification from immersion in the control reaction solution. While the methyl region of the IR spectrum provides evidence of the bound peptide on the surface, the amide region (1400−1800 cm−1) provides additional information regarding the secondary structure of the peptides bound to the surface.11,45−48 Figure 2 shows representative infrared spectra of the amide region containing peptide backbone stretches of the carbonyl group (amide I, centered around 1650 cm−1) and stretching of the C−N bond coupled with N−H bending (amide II, near 1545 cm−1). A spectrum of the azide-terminated surface before reaction (Figure 2, green) lacked any peaks in 3443
DOI: 10.1021/la5049369 Langmuir 2015, 31, 3441−3450
Article
Langmuir
sheets in an antiparallel configuration and arises from transition dipole coupling of carbonyl groups along the peptide backbone.45,46,48−50 The less-intense peaks at 1666 and 1675 cm−1 were attributed to random or β-turn conformations.45,51,52 Similar to the 18 h reaction, the 30 min reacted sample (Figure 2, blue) also showed splitting of the amide I peak at 1637 and 1694 cm−1 along with random and β-turn absorbances at 1662 and 1679 cm−1. The signal-to-noise ratio of this sample is lower than the 18 h reaction due to low peptide surface coverage and represents the shortest reaction time point that allows detection by IR spectroscopy. Overall, the increased methyl and methylene peaks, as well as the presence of amide I and II stretches in the reacted sample spectra, provide evidence of both the presence and the structure of surface-bound peptides. The splitting of the amide I peak indicates that the peptides are self-assembled into a β-sheet conformation in an antiparallel alignment. The less intense stretches in the amide I region attributed to random or β-turn structures could arise either from peptides adopting conformations that stabilize the β-sheets or because the SAM surfaces were 100% functionalized in reactive azide groups, and it was expected that some peptides could bind to the surface in orientations that do not facilitate the formation of organized β-sheets. Spectra of the control samples showed only very small increases in methyl/methylene signals and negligible absorbance in the amide region, indicating that peptides only bind to the surface in the presence of the Cu(I) catalyst in the reaction solution. Collectively, the FTIR data demonstrate the presence of surface-bound peptide in a largely antiparallel βsheet conformation, which requires the Huisgen Cu(I) catalyst to form the triazole linkage that tethers the peptide to the underlying SAM. In addition to the methyl, methylene, and amide regions, the azide peak (2103 cm−1) of the SAM functionalized surfaces was also monitored to confirm SAM formation, but peak degradation was not used for quantitative analysis due to decomposition pathways unrelated to the Huisgen reaction. In addition to vibrational spectroscopy, XPS was also used to confirm the presence of the surface-bound peptide. Figure 3 shows representative spectra of the C 1s region from an 18 h β6-reacted sample and an unreacted azide-terminated SAM, including component fits. The C 1s signal from the unreacted azide surface (Figure 3, bottom) showed peaks corresponding to C−C (284.5 eV) and C−N (285.8 eV) bonds, representing the alkyl chains of the SAM as well as the C−N bonds of the terminal azide groups. The β6-reacted surface (Figure 3, top) contained peaks arising from photoelectron emission from C− C, C−(N,O), and CO bonds near 284.5, 285.8, and 287.7 eV, respectively.53 The development of the carbonyl peak as well as the increase in detected C−(N,O) photoelectrons on the β6-reacted samples relative to the unreacted surface are consistent with the presence of amide groups on the surface. These results indicate the successful attachment of peptides to the gold surface and are consistent with previous XPS analysis of surface-associated peptides.4,41 Structural Characterization of Surface-Bound Peptides. CD spectra were collected to identify any differences in secondary structure that the peptide adopts in solution and on the surface. Figure 4 shows the measured ellipticity of the β6 peptide in HPW solvent (red), with a negative band
