A Genetic Approach for Controlling the Binding and Orientation of

Jan 15, 2008 - E. James Davis, and François Baneyx*. Department of Chemical Engineering, UniVersity of Washington, Seattle, Washington 98195-1750...
0 downloads 0 Views 367KB Size
2000

Langmuir 2008, 24, 2000-2008

A Genetic Approach for Controlling the Binding and Orientation of Proteins on Nanoparticles Atanu Sengupta,† Corrine K. Thai, M. S. R. Sastry,‡ James F. Matthaei, Daniel T. Schwartz, E. James Davis, and Franc¸ ois Baneyx* Department of Chemical Engineering, UniVersity of Washington, Seattle, Washington 98195-1750 ReceiVed July 11, 2007. In Final Form: NoVember 7, 2007 Although silver nanoparticles are excellent surface enhancers for Raman spectroscopy, their use to probe the conformation of large proteins at interfaces has been complicated by the fact that many polypeptides adsorb weakly or with a random orientation to colloidal silver. To address these limitations, we sought to increase binding affinity and control protein orientation by fusing a silver-binding dodecapeptide termed Ag4 to the C-terminus of maltosebinding protein (MBP), a well-characterized model protein with little intrinsic silver binding affinity. Quartz crystal microbalance measurements conducted with the MBP-Ag4 fusion protein revealed that its affinity for silver (Kd ≈ 180 nM) was at least 1 order of magnitude higher than a control protein, MBP2, containing a non-silver-specific C-terminal extension. Under our experimental conditions, MBP-Ag4 SERS spectra exhibited 2-4 fold higher signalto-background relative to MPB2 and contained a number of amino acid-assigned vibrational modes that were either weak or absent in control experiments performed with MBP2. Changes in amino acid-assigned peaks before and after MBP-Ag4 bound maltose were used to assess protein orientation on the surface of silver nanoparticles. The genetic route described here may prove useful to study the orientation of other proteins on a variety of SERS-active surfaces, to improve biosensors performance, and to control functional nanobiomaterials assembly.

Introduction Since the discovery of intense Raman bands from pyridine adsorbed on an anodized silver electrode by Fleischmann and co-workers1 and the pioneering work of Jeanmaire and Van Duyne,2 surface-enhanced Raman spectroscopy (SERS) has emerged as a powerful tool to probe the structure of macromolecules adsorbed on the surface of nanoscale metals such as silver and gold. The remarkable (103-1014) increase in the intensity of inelastically scattered light observed with SERS has been attributed to two mechanisms. The first is an optical effect. For a metal, and at certain frequencies of incident irradiation, absorption and scattering cross-sections become very large (i.e., there is a resonance between the incident photon frequency and the oscillation of the conduction electrons in the metal). Furthermore, the internal electric field becomes localized near the surface of a metallic particle (the localized surface plasmon resonance or LSPR).3 The second enhancement mechanism is chemical and has been assigned to charge-transfer or bond formation between metal and adsorbate.4 SERS is suitable for the characterization of aqueous samples, requires small amounts of materials, and is sensitive enough for single-molecule studies.5,6 Because the illumination wavelength * To whom correspondence should be addressed. E-mail: baneyx@ u.washington.edu. † Current address: Real-time Analyzers, Inc., 362 Industrial Park Road, CT 06457. ‡ Current address: Division of Molecular Biology and Biochemistry, School of Biological Sciences, University of Missouri, Kansas City, MO 64110. (1) Fleischmann, M.; Hendra, P. J.; McQuillam, A. J. Raman-spectra of pyridine adsorbed at a silver electrode. Chem. Phys. Lett. 1974, 26, 163. (2) Jeanmaire, D. L.; Van Duyne, R. P. Surface Raman spectro-electrochemistry. Part I. Heterocyclic, aromatic and aliphatic amines adsorbed on the ionized silver electrode. J. Electroanal. Chem. Interfacial Electrochem. 1977, 84, 1. (3) Stuart, D. S.; Haes, A. J.; Yonzon, C. R.; Hicks, E. M.; Van Duyne, R. P. Biological applications of localized surface plasmonic phenomena. IEE Proc. Nanobiotechnol. 2005, 152, 13. (4) Moskovits, M. Surface-enhanced Raman spectroscopy: a brief retrospective. J. Raman Spectroscopy 2005, 36, 485.

can be selected to minimize fluorescence and sample damage and because it yields a vibrational fingerprint of the first few molecular layers of adsorbed species, SERS has found widespread applications in biosensing and chemical detection.7,8 SERS can also provide information on the orientation of amino acids and short peptides at solid surfaces.9-13 Although a detailed knowledge of how larger proteins interact with solids would be valuable to optimize sensor performance and to tailor the assembly of complex structures in nanobiotechnology and tissue engineering, few SERS studies have tackled the problem of protein orientation at interfaces, and most have focused on robust polypeptides containing an easily detectable chromophore.14-17 Lack of progress in this area is in part due to the facts that many proteins exhibit a weak affinity for silver (an excellent (5) Kneipp, K.; Wang, Y.; Kneipp, H.; Perelman, L. T.; Ltzkan, L.; Dasari, R. R.; Feld, M. S. Single molecule detection using surface-enhanced Raman scattering (SERS). Phys. ReV. Lett. 1997, 78, 1667. (6) Nie, S.; Emory, S. R. Probing single molecules and single nanoparticles by surface-enhanced Raman, scattering. Science 1997, 275, 1102. (7) Aroca, R. F.; Alvarez-Puebla, R. A.; Pieczonka, N.; Sanchez-Cortez, S.; Garcia-Ramos, J. V. Surface-enhanced Raman scattering on colloidal nanostructures. AdV. Colloids Interface Sci. 2005, 116, 45. (8) Willets, K. A.; Van Duyne, R. P. Localized surface plasmon resonance spectroscopy and sensing. Annu. ReV. Phys. Chem. 2007, 58, 267. (9) Herne, T. M.; Ahern, a. M.; Garrell, R. L. Surface-Enhanced RamanSpectroscopy of Peptides - Preferential N-Terminal Adsorption on Colloidal Silver. J. Am. Chem. Soc. 1991, 113, 846. (10) Podstawka, E.; Ozaki, Y.; Proniewicz, L. M. Part III: surface-enhanced Raman scattering of amino acids and their homodipeptide monolayers deposited onto colloidal gold surface. Appl. Spectrosc. 2005, 59, 1516. (11) Podstawka, E.; Sikorska, E.; Proniewicz, L. M.; Lammek, B. Raman and surface-enhanced Raman spectroscopy investigation of vasopressin analogues containing 1-aminocyclohexane-1-carboxylic acid residue. Biopolymers 2006, 83, 193. (12) Singha, a.; Dasgupta, S.; Roy, A. Comparison of metal-amino acid interaction in Phe-Ag and Tyr-Ag complexes by spectroscopic measurements. Biophys. Chem. 2006, 120, 215. (13) Stewart, S.; Fredericks, P. M. Surface-enhanced Raman spectroscopy of peptides and proteins adsorbed on an electrochemically prepared silver surface. Spectrochim. Acta Part A: Molec. Biomolec. Spectrosc. 1999, 59, 1516. (14) Ahem, A. M.; Garrell, R. L. Protein-metal interactions in protein-colloid conjugates probed by surface-enhanced Raman spectroscopy. Langmuir 1991, 7, 254.

10.1021/la702079e CCC: $40.75 © 2008 American Chemical Society Published on Web 01/15/2008

Binding and Orientation of Proteins on Nanoparticles

SERS enhancer),10,16 that native proteins adopt a random orientation following adsorption on the metal, and that they may progressively change conformation or denature at surfaces.14 In this study, we show that the above limitations can be alleviated by fusing a short silver-binding peptide, termed Ag4,18 to a permissive site of the protein under study (i.e., a location at which an extraneous stretch of amino acids does not affect protein folding, structure, or function).19,20 By increasing the affinity of the target polypeptide for colloidal silver, specifying protein orientation with respect to the metallic surface and acting as a spacer, the Ag4 moiety leads to more intense and informationrich SERS spectra. Using maltose-binding protein (MBP) as a model system, we further show that it is possible to combine SERS data and crystal structure information to infer the orientation of an MBP-Ag4 fusion protein on colloidal silver nanoparticles. Materials and Methods Plasmid Construction. Plasmid pMal-c2 (New England Biolabs), which encodes a signal sequence-less version of the MBP followed by a multiple cloning site and the lacZR peptide, was digested with SacI and XbaI. The free DNA ends were dephosphorylated with shrimp alkaline phosphatase (Fermentas). The large fragment was recovered by low-melting-point agarose electrophoresis and purified using the Qiagen PCR purification kit. Two phosphorylated primers encoding a SSSGGG flexible linker followed by a codon-optimized version of the Ag4 peptide18 and two stop codons together with extensions compatible with the overhangs generated upon SacI and XbaI digestion (5′-CGGGCGGTGGCAACCCGAGCAGCCTGTTTCGCTATCTGCCGAGCGATTGATGAT-3′ and 5′-CTAGATCATCAATCGCTCGGCAGATAGCGAAACAGGCTGCTCGGGTTGCCACCGCCCGAGCT-3′)werepurchased from Invitrogen. Primers were resuspended in doubly distilled H2O (ddH2O) to a final concentration of 200 µM. Aliquots (10 µL) of each primer were mixed with 10 µL of 2X annealing buffer (80 mM Tris-HCl, pH 8.0, 20 mM MgCl2, 100 mM NaCl), and the two solutions were combined. The mixture was transferred to a heating block, incubated at 99 °C for 10 min, and allowed to cool to room temperature by turning off the heat. The annealed DNA fragment was purified using the Centri-Sep kit (Princeton Separations), mixed with the pMal-c2 backbone, and ligated. Top10 (Invitrogen) transformants were screened for insert-containing plasmids. One such clone was selected, and the sequence was verified by DNA sequencing using primer 5′-CCACGTATTGCCGCCACCATGGAAAACGCCCAG-3′. The plasmid was named pMalE-Ag4. Protein Purification. Top10 cells harboring pMalE-Ag4 were grown at 37 °C in 500 mL of LB medium containing 0.2% glucose and 50 µg/mL carbenicillin to mid-exponential phase (A600 ≈ 0.5). Cultures were supplemented with 350 µM of isopropyl-β-Dthiogalactoside (IPTG) to induce transcription from the tac promoter, and the fusion protein was allowed to accumulate for 3.5 h. Cells were harvested by centrifugation at 5000g for 10 min, resuspended into 15 mL of running buffer (20 mM Tris-HCl pH 7.4, 200 mM NaCl, 1 mM EDTA), and disrupted via three cycles on a French Press operated at 10 000 psi. Following centrifugation at 10 000g for 10 min to remove insoluble material, clarified extracts were (15) Delfino, I.; Bizzarri, A. R.; Cannistraro, S. Single-molecule detection of yeast cytochrome c by surface-enhanced Raman spectroscopy. Biophys. Chem. 2005, 113, 41. (16) Keating, C. D.; Kovaleski, K. M.; Natan, M. J. Protein:colloid conjugates for surface-enhanced Raman scattering: stability and control of protein orientation. J. Phys. Chem. B 1998, 102, 9409. (17) Picorel, R.; Chumanov, G.; Cotton, T. M.; Montoya, G.; Toon, S.; Seibert, M. Surface-Enhanced Resonance Raman-Scattering Spectroscopy of PhotosystemIi Pigment-Protein Complexes. J. Phys. Chem. B 1994, 98, 6017. (18) Naik, R. R.; Stringer, S. J.; Agarwal, G.; Jones, S. E.; Stone, M. O. Biomimetic synthesis and patterning of silver nanoparticles. Nat. Mater. 2002, 1, 169. (19) Manoil, C.; Bailey, J. A simple screen for permissive sites in proteins: analysis of the Escherichia coli lac permease. J. Mol. Biol. 1997, 267, 250. (20) Manoil, C.; Traxler, B. Insertion of in-frame sequence tags into proteins using transposons. Methods 2000, 20, 55.

Langmuir, Vol. 24, No. 5, 2008 2001 loaded onto a 15 mL amylose-agarose column (New England Biolabs). The column was developed in running buffer at a flow rate of 1 mL/min. Unbound proteins were removed by passing five column volumes of running buffer through the column, and MBP-Ag4 was eluted by addition of 20 mM maltose to the column buffer. For all experiments, aliquots from MBP-Ag4 stocks (>95% purity as judged by SDS-PAGE analysis) and MBP2 stocks (purchased from New England Biolabs) were dialyzed twice for 15 h against 2 L of ddH2O and concentrated with microconcentrators (Amicon), and final protein concentrations were determined using the Coomassie dye-binding protein assay kit (Sigma). MBP-Ag4 purified in this fashion bound quantitatively to amylose-agarose, suggesting that it was mostly maltose-free. Quartz Crystal Microbalance (QCM) Measurements. All measurements were conducted on an ELCHEMA EQCN-700 electrochemical QCM with 10.0 MHz AT-cut crystals coated with a 100 nm silver film (International Crystal Manufacturing) as previously described21 except that the crystals were immersed in 3 mL of ddH2O. For the experiments of Figure 4b, MBP-Ag4 was injected at a final concentration of 2Kd (360 nM). After 75 min, maltose (Sigma) was injected at a final concentration of 100 µM. Preparation and Characterization of Silver Nanoparticles. Colloidal silver was produced using the method of Keir et al.22 Briefly, 9 mL of an aqueous solution of 2.2 mM AgNO3 was added dropwise and with continuous stirring to 75 mL of 1.2 mM NaBH4 in a conical flask held on ice. At the end of this step, the flask was removed from the ice bath and stirring was continued for an additional 45 min. UV/visible spectra were collected on a Beckman Coulter DU640 spectrophotometer 15 min after supplementing 1.8 mL of colloidal silver suspension with 200 µL of ddH2O or with 200 µL of MBP-Ag4 or NaCl stock solutions to final concentrations of 5.75 and 200 mM, respectively. Photon correlation spectroscopy experiments were performed on the same samples using a Brookhaven ZetaPALS instrument. Raman Spectroscopy. The Raman setup used in this study has been described elsewhere.23 Briefly, a Coherent Innova 300 continuous-wave laser operating at a wavelength of 514.5 nm was used to illuminate an optical glass cuvette (1 × 1 × 4.5 cm3) containing 3 mL of analyte and silver colloid. The scattered light was focused on the slit of a 500 mm focal length single pass Acton SpectraPro-500I monochromator with a grating turret permitting dispersions of 300, 1200, and 2400 lines/mm. A SuperNotch-Plus holographic filter (Kaiser Optical Systems) was placed in front of the spectrometer to remove the laser line, and spectra were recorded using a Princeton Instruments back-illuminated, liquid-nitrogencooled 1340 × 100 pixel array CCD camera. The laser beam power at the cuvette was less than 100 mW, and the exposure and integration time was 120 s. The background librational contribution of water was subtracted from all spectra except for that of Figure 5b in which major Raman bands are easily identified. Because SERS signal enhancement is sensitive to the ratio of bioanalyte to silver colloid,24 we first optimized this parameter by recording the spectra of MBP-Ag4 at various concentrations in the same silver colloid preparation (data not shown). Comparison of the 963 cm-1 peak normalized to the 1665 cm-1 peak (a combination of the water scissoring mode peak at 1635-1645 cm-1 and of the amide I band of amino acids at 1655-1675 cm-1)25 showed that the (21) Dai, H.; Choe, W. S.; Thai, C. K.; Sarikaya, M.; Traxler, B. A.; Baneyx, F.; Schwartz, D. T. Nonequilibrium synthesis and assembly of hybrid inorganicprotein nanostructures using an engineered DNA binding protein. J. Am. Chem. Soc. 2005, 127, 15637. (22) Keir, R.; Sadler, D.; Smith, W. Preparation of stable, reproducible silver colloids for use as surface-enhanced resonance Raman scattering substrates. Appl. Spectrosc. 2002, 56, 551. (23) Sengupta, A.; Laucks, M. L.; Dildine, N.; Drapala, E.; Davis, E. J. Bioaerosol characterization by surface-enhanced Raman spectroscopy (SERS). J. Aerosol Sci. 2005, 36, 651. (24) Sengupta, A.; Laucks, M. L.; Davis, E. J. Surface-enhanced Raman spectroscopy of bacteria and pollen. Appl. Spectrosc. 2005, 59, 36. (25) Mikkelsen, R. B.; Verma, S. P.; Wallach, D. F. H. Effect of transmembrane ion gradient on Raman spectra of sealed, hemoglobin-free eythrocyte membrane vesicles. Proc. Natl. Acad. Sci. U.S.A. 1978, 75, 5478.

2002 Langmuir, Vol. 24, No. 5, 2008

Figure 1. The Ag4 extension increases the affinity of maltosebinding protein for silver. (a) C-terminus amino acid composition of native maltose-binding protein (MBP), its silver-binding variant (MBP-Ag4) and of the negative control MBP2. In the MBP-Ag4 sequence, the flexible linker is in blue and the silver-binding dodecapeptide in red. (b) Langmuir adsorption isotherms of MBPAg4 (filled symbols) and MBP2 (open symbols) on polycrystalline silver were constructed from QCM data. Deviations between replicate measurements were less than 10%. The inset is a SDS polyacrylamide gel showing crude cell lysate (lane 1) and purified MBP-Ag4 (lane 2). Lane M contains molecular mass markers. signal intensity increased monotonously from 1 to 5.75 µM but decreased at MBP-Ag4 concentrations higher than 6.6 µM. Consequently, a protein concentration of 5.75 µM was selected for all subsequent experiments. Structure Rendering. Ribbon structures and protein molecular surfaces were calculated and rendered using the SwissPdb Viewer software26 and the crystal structure coordinates of unliganded MBP27 (Protein data bank accession 1JW4) and maltose-bound MBP28 (Protein data bank accession number 1ANF).

Results and Discussion Construction and Characterization of a Silver-Binding Derivative of the MBP. To endow MBP with silver-binding ability, we used standard molecular biology techniques to fuse the Ag4 dodecapeptide (NPSSLFRYLPSD), which was previously isolated as a silver-binder by bacteriophage M13 display,18 to the C-terminal Thr-366 of mature MBP. The flexible sequence SSSGGG was engineered as a fusion joint to separate the Ag4 peptide from the MBP framework. The resulting chimaera, called MBP-Ag4 (Figure 1a), was expressed at high level in the cytoplasm of Escherichia coli and purified to near-homogeneity by a single affinity chromatography step on a cross-linked amylose, indicating that the fusion protein retained maltose binding ability (Figure 1b, inset). The affinity of MBP-Ag4 for polycrystalline silver was quantified by QCM measurements. In this technique, protein adsorption to an Ag-coated quartz crystal leads to a progressive decrease in resonant frequency that reaches a stable minimum (∆f) at equilibrium. At high protein concentration (c), saturation of adsorption occurs. This corresponds to a maximum resonant (26) Guex, N.; Peitsch, M. C. SWISS-MODEL and the Swiss-Pdb Viewer: an environment for comparative protein modeling. Electrophoresis 1997, 18, 2714. (27) Duan, X.; Quiocho, F. A. Structural evidence for a dominant role of nonpolar interactions in the binding of a transport/chemosensory protein to its highly polar ligands. Biochemistry 2002, 41, 706. (28) Quiocho, F. A.; Spurlino, J. C.; Rodseth, L. E. Extensive features of tight oligosaccharide binding revealed in high-resolution structures of the maltodextrin transport/chemosensory receptor. Structure 1997, 5, 997.

Sengupta et al.

frequency shift denoted by ∆fmax. By plotting QCM data collected at various concentrations under the form of a Langmuir adsorption isotherm (∆f ) [∆fmaxc]/[Kd + c]), one can readily extract the equilibrium dissociation constant (Kd) which corresponds to the protein concentration at which the silver crystal is covered with a half-monolayer of protein. In the case of MBP-Ag4, this exercise led to a ∆fmax of ≈90 Hz and a Kd of 180 nM (Figure 1b, b). Because proteins spontaneously adsorb on high-surface-energy noble metals such as silver and gold, we repeated QCM measurements with MBP2. This commercially available derivative of the maltose-binding protein contains a C-terminal extension that is similar in sizesbut not in compositionsto the Ag4 peptide (Figure 1a). It is, therefore, not expected to confer any particular silver binding affinity to MBP since it was not selected for this activity. Using the initial kinetics data of Figure 1b (O) and a ∆fmax of 90 Hz, we calculated a Kd value for MBP2 on silver of 1.9 µM. The true equilibrium dissociation constant is likely to be higher since MBP2 should be more randomly orientedsand therefore less tightly packed on the silver surfacesthan MBPAg4. We conclude that fusion of the Ag4 dodecapeptide to the C-terminus of MBP enhances its affinity for silver by at least one order of magnitude. To determine if the adsorption of MBP-Ag4 to silver nanoparticles would affect their optical properties and/or morphology, we produced colloidal silver by reduction of silver nitrate with sodium borohydride and collected UV/visible absorption spectra following addition of ddH2O or MBP-Ag4. Figure 2a shows that the pure colloid exhibited a narrow peak at 390 nm that is characteristic of the surface plasmon absorption of silver nanoparticles (black spectrum). Addition of MBPAg4 led to the appearance of a 280 nm peak corresponding to the absorption band of the proteins’ Trp (W) and Tyr (Y) residues and caused a 9 nm red-shift in the silver plasmon peak, as well as a small decrease in its intensity (Figure 2a, red spectrum). By comparison, when colloid aggregation was induced by addition of 200 mM NaCl, the silver plasmon peak disappeared entirely (Figure 2a, blue spectrum). We next performed photon correlation spectroscopy (PCS) experiments on the same samples. Figure 2b shows that, whereas the light-scattering autocorrelation functions of the pure and MBP-Ag4-supplemented colloids were similar, that of the NaCl-treated sample exhibited a significant increase in time scale decay. The average hydrodynamic diameters extracted from PCS data were 17 nm for colloidal silver, 19 nm for the MBP-Ag4-treated sample, and 233 nm for the NaCltreated sample. Thus, while adsorption of an MBP-Ag4 protein shell on the surface of silver nanoparticles leads to slight changes in their optical properties, it does not cause them to aggregate. The Silver-Binding Domain Increases the Information Content of SERS Spectra. As schematically depicted in Figure 3a, we hypothesized that the presence of the silver-binding moiety would improve the intensity and quality of MBP SERS spectra both by increasing the number of protein molecules bound per silver particle and by minimizing orientation variations compared to random adsorption. To test this idea, MBP-Ag4 (or MBP2 as a control) was mixed with colloidal silver at a 5.75 µM final concentration, a value that we found to be optimal for data acquisition under our experimental conditions (see Materials and Methods, Raman Spectroscopy). Because this concentration exceeds the Kd of both MBP-Ag4 and MBP2, the nanoparticle surface should have a significant layer of adsorbed protein, a prediction consistent with the 2 nm increase in average hydrodynamic diameter between MBP-Ag4-treated and pure colloids (Figure 2b).

Binding and Orientation of Proteins on Nanoparticles

Figure 2. Influence of MBP-Ag4 on the absorption characteristics and size of silver nanoparticles. (a) UV/visible absorption spectra of colloidal silver supplemented with ddH2O, MBP-Ag4 to a 5.75 µM final concentration, or NaCl to a 200 mM final concentration. (b) Autocorrelation functions of scattered light from samples of colloidal silver supplemented with ddH2O, MBP-Ag4 to a 5.75 µM final concentration, or NaCl to a 200 mM final concentration. All data were recorded after 15 min incubation at room temperature.

As expected from the fact that MBP-Ag4 and MBP2 have 371 amino acids in common and only differ over ∼4% of their sequence, the two spectra (corrected for the librational background of water and normalized to the 1665 cm-1 peak to allow for direct comparison) exhibited very similar features (Figure 3b). Nevertheless, the signal-to-background ratio was 2-4-fold higher for MBP-Ag4 compared to MBP2, an anticipated trend consistent with the fact that the Ag4 moiety should lead to preferential orientation of MBP-Ag4 on silver. In addition, the MBP-Ag4 spectrum contained a number of peaks that were either undetectable or weak in the MBP2 spectrum (Figure 3b, dashed lines and Table 1, italic entries). Using data from normal Raman and SERS studies of individual amino acids,29-32 we assigned several of these enhanced vibration modes to specific amino acids (Table 1). Strong signatures were (29) Guicheteau, J.; Argue, L.; Hyre, A.; Jacobson, M.; Christesen, S. D. Raman and surface-enhanced Raman spectrsocopy of amino acids and nucleotide bases for target bacterial vibrational mode identification. Proc. SPIE 2006, 6218, 62180O. (30) Kim, S. K.; Kim, M. S.; Suh, S. W. Surface-enhanced Raman scattering (SERS) of aromatic amino acids and their glycyl dipeptides in silver sol. J. Raman Spectrosc. 1987, 18, 171. (31) O’Neal, P. D.; Cote, G. L.; Motamedi, M.; Chen, J.; Lin, W. C. Feasibility study of using surface-enhanced Raman spectroscopy for the quantitative detection of excitatory amino acids. J. Biomed. Optics 2003, 8, 33.

Langmuir, Vol. 24, No. 5, 2008 2003

Figure 3. The Ag4 extension enhances the intensity and richness of the maltose-binding protein SERS spectrum. (a) Schematic representation of the adsorption of MBP2 (left) and MBP-Ag4 (right) onto colloidal silver. (b) SERS spectra of MBP-Ag4 (red) and MBP2 (black). The protein concentration was 5.75 µM. Vibrational modes that are enhanced in the MBP-Ag4 spectrum and the 1665 cm-1 normalization peak are indicated by dashed lines and labeled. Spectra were collected at 100 mW power and at 514 nm with an integration time of 120 s.

identified for F, W, and G, moderate ones for Y, S, D, and E, and weaker ones for T, K, P, L, V, and N. Among the above 13 residues, four (F, Y, P, D) are present in the Ag4 dodecapeptide but not in the C-terminal extension of MBP2 (Figure 1a). Because the Ag4 domain should be in close proximity to the silver surface where surface-enhancement is maximal, it is likely that the F, Y, P, and D peaks include large contributions from the silverbinding peptide. However, because it contains 15 Y, 15 F, 21 P, and 24 D residues, MBP itself is likely to contribute to the intensity increase in these vibrational modes. Indeed, peaks corresponding to amino acids that are absent in the Ag4 peptide (e.g., W, E, K, V, and T) are enhanced in the MBP-Ag4 spectrum, presumably because Ag4-mediated tethering brings the W, E, K, V and T residues of maltose-binding protein closer to the nanoparticle surface and/or specifies the orientation of MBPAg4 compared to a more randomly adsorbed MBP2. Probing MBP-Ag4 Conformational Changes with SERS. Upon maltose binding, the two structural lobes of the maltosebinding protein rotate by ≈35° and twist laterally by ≈8° relative to each other (Figure 4a).33 This major conformational change leads to the closing of the structure around a hinge region and a 7 Å migration of the N- and C-termini toward each other. Because surface-enhancement rapidly decays with the distance (32) Podstawka, E.; Ozaki, Y.; Proniewicz, L. M. Part I: Surface-enhanced Raman spectroscopy investigation of amino acids and their homodipeptides adsorbed on colloidal silver. Appl. Spectrosc. 2004, 58, 570. (33) Sharff, A. J.; Rodseth, L. E.; Spurlino, J. C.; Quiocho, F. A. Crystallographic evidence of a large ligand-induced hinge-twist motion between the two domains od the maltodextrin-binding protein involved in active transport and chemotaxis. Biochemistry 1992, 31, 10657.

2004 Langmuir, Vol. 24, No. 5, 2008

Sengupta et al.

Table 1. Tentative Peak Assignments for the Vibration Modes Contributed by MBP-Ag4 Residues before and after Maltose Addition frequency (cm-1) no maltosea 594 620 (m)

635 (m)

681(vw, sh) 721 (vw, sh) 752

812 (w) 826 (w)

860

890

906 (w) 963 1004 (w)

1028

1051

1105

1133 (m)

tentative assignmentb CONH2 vibration for N (SERS 610 cm-1 band) side chain -CdC- bend and ring deformation of monosubstituted benzenes for F (NR 620 cm-1 band) side chain -CdC- bend and ring deformation of monosubstituted benzenes for F (SERS 634 cm-1 band) and Y (NR 642 cm-1 band) COO- wagging for V (NR 664 cm-1 band) ? benzene stretch and pyrrole in-phase breathing for W (NR 755 cm-1 and SERS 758 cm-1 bands) CH2 and CH3 rocking and C-C stretch for V (NR 753 cm-1 band) C-C stretch for Y Y doublet, phenol ring breathing and para-disubstituted benzenes for Y (NR 829 cm-1 band) C-C skeletal stretch and C-CH3 stretch for F (SERS 832 cm-1 band) C-C-N stretch for S (NR 869 cm-1 band) and T (SERS 865 cm-1 band) C-C skeletal stretch and C-CH3 stretch for D (SERS 862 cm-1 band) N-H bending and indole ring vibration for W (NR 873 cm-1 and SERS 877 cm-1 bands) C-C skeletal stretch and C-COO stretching for G (SERS 891 cm-1 band) C-N stretch for R C-C and C-COOH stretch for G (NR 915 cm-1 band) C-C stretch and C-N stretch for L (SERS 951 cm-1 band) aromatic ring breathing mode for F (SERS 1001 cm-1 band) C-C stretch, C-N stretch and benzene/pyrrole out-of-phase breathing for W (NR 1008 cm-1 band) in-plane CH deformation (monosubstituted benzene) for F (SERS 1030 cm-1 band) CH2 wagging for P (SERS 1027 cm-1 band) C-C stretch for Y, G, K C-NH2, C-N, and C-C stretch for D , Y, S, A, and V (SERS 1045, 1046, 1047, 1045, 1054, and 1056 cm-1 bands, respectively) C-N and C-NH3 stretch for A (NR 1112 cm-1 band) C-C stretch, NH2 twist and NH3 wagging for V (NR 1125 cm-1 band) C-C stretch, NH2 twist, and NH3 wagging for L (NR 1130 cm-1 band) C-C stretch for Y and V

probable amino acidc

frequency (cm-1) with maltosea

N

599

F

none

F, Y

635

closest V

688

? W, V

none 747

Y? Y, F

none none

S, T, D

862

W

G

892

R? G

911 (vw)

L

963

F, W

none

F, P

none

Y? G? K? D, Y, S, A, V

1051

A

1100

V

L

Y? V?

none

Binding and Orientation of Proteins on Nanoparticles

Langmuir, Vol. 24, No. 5, 2008 2005

Table 1. (Continued) frequency (cm-1) no maltosea 1154 (sh)

1167

1194, 1213 (w)

1230

1252 (w)

1294

1339, 1345

1371, 1391

1430

1482

tentative assignmentb C-C stretch, NH2 twist and NH3 wagging for G (SERS 1152 cm-1 band) C-C stretch for I and V CH-NH2 for Y (SERS 1166 cm-1 band) C-C stretch (aromatic residue) for F (NR 1180 cm-1 band) C6H5-C vibration (alkyl benzene) and phenyl C stretch for F (SERS 1202 cm-1 and NR 1181 and 1201 cm-1 bands) CH2 twist and rock, NH3 rocking, and CH-NH2 twist and rock for F (NR 1181 cm-1 band) and V (NR 1191 cm-1 band) CH2 twist and rock and CH2 vibration for V (SERS 1213 cm-1 band) CH2 twist and rock and CH2 vibration for E (SERS 1236 cm-1 band) amide III, N-H in plane bend, C-N stretch for Y CH2 twist and rock for L, E, and I (NR 1242, 1258, and 1259 cm-1 bands, respectively) CH2 twist and rock and CH2 wagging for G, S, E, and A (NR 1300, 1301, 1301, and 1304 cm-1 bands, respectively), and V (SERS 1296 and 1311 cm-1 bands) amide III, N-H in-plane bend, CH-CH2 deformation and C-C stretch for K and Y doublet in proteins and Fermi resonance between N-C in pyrrole ring for W (NR 1339 and 1357 cm-1 bands) C-H deformation and COOsymmetric stretch for L (SERS 1333 cm-1 band) C-H bending and vibration and C-H stretch for V (SERS 1354 cm-1 band) ring and C-H vibration for Y (NR 1327 cm-1 band) CH2 twist or rock for S and G ring stretch for F (SERS 1378w cm-1 band) CH3 symmetrical deformation, COO- symmetrical stretch, C-NH3+ stretch, and CRH2 wagging for G and S (SERS 1383w and 1401 cm-1 bands, respectively) C-N stretch, COO- vibration, and COO- symmetrical stretch for N, E, and D (SERS 1396, 1397 and 1398 cm-1 bands, respectively) COO- symmetric stretch and CH2 deformation for W (SERS 1423 cm-1 band) COO- symmetric stretch for L and G (SERS 1424 and 1425 cm-1 bands, respectively) CH, CH2, and CH3 deformation for S and Y (SERS 1479 and 1482 cm-1 bands, respectively)

probable amino acidc

frequency (cm-1) with maltosea

G

none

I? V? Y, F

1170

V, F

1194 (w), 1213 (vw)

E

1230 (vw)

Y? L, E, I

1256 (w)

G, S, E, A, V

1297

K? Y?

W, V, L, Y

1346

S? G?

F

none

G, S, N, E, D

W, L, D

1430

Y, S

1485

2006 Langmuir, Vol. 24, No. 5, 2008

Sengupta et al.

Table 1. (Continued) frequency (cm-1) no maltosea 1505, 1530

1655 (sh), 1665

tentative assignmentb C-C stretching and pyrrole ring for W (SERS 1549 cm-1 band) COO- antisymmetric vibration for V (SERS 1543 cm-1 band) overlap of HOH scissor and amide I

probable amino acidc

frequency (cm-1) with maltosea

W, V

1510 (w), 1530 (vw, sh)

all

1645 (sh), 1665

a

Abbreviations are: w, weak; vw, very weak; m, medium; sh, shoulder. Bands in italics are either absent or weak in the MBP2 spectrum. b Vibration modes from normal Raman (NR) or SERS studies of individual amino acids that fall in the vicinity of the experimental MBP-Ag4 Raman shifts are indicated. c Amino acids are identified by their one-letter code. Residues in bold exhibit vibration modes within (10 cm-1 of experimental peaks and are thus assigned with higher confidence. Assignments made with lower confidence are in plain text or followed by a question mark. Underlined amino acids correspond to vibration modes that become weak or undetectable following maltose addition.

Figure 4. Maltose addition does not significantly affect the affinity of MBP-Ag4 for polycrystalline silver. (a) Ribbon structures of MBP in its maltose-free (left) and maltose-bound (right) forms. The amino (N) and carboxyl (C) termini are indicated, and maltose is shown in yellow in the liganded structure. (b) QCM trace showing the frequency decrease accompanying the binding of MBP-Ag4 (260 nM) to the silver crystal. At 75 min, maltose was added to a final concentration of 100 µM (arrow).

to the surface,34 SERS primarily probes the vibrational properties of the first 1-2 nm of adsorbed species. Given the dimensions of maltose-binding protein (≈3 × 4 × 6.5 nm3)33,35 and the fact that Ag4-mediated immobilization should orient the protein on the silver surface, we reasoned that comparison of MBP-Ag4 SERS spectra collected with or without maltose might provide valuable information on the conformational changes induced by ligand binding and on how the protein interacts with silver. We first verified that binding of maltose to the active site of MBP-Ag4 would not cause its release from the noble metal surface. The surface of the silver-coated QCM crystal is polycrystalline and therefore displays any combination of crystalline planes that might be found on a single-crystal silver nanoparticle. Thus, monitoring the frequency decrease accompanying the adsorption of MBP-Ag4 by microbalance experiments (Figure 4b; t ) 25 min), as well as any subsequent changes upon maltose binding (t ) 75 min), provides a window into processes occurring on the nanoparticles. Specifically, Figure 4b shows that injection of a nearly 400-fold molar excess of maltose once protein adsorption is complete leads to a small (34) McCall, S. L.; Platzman, P. M. Raman-scattering from chemisorbed molecules at surfaces. Phys. ReV. B 1980, 22, 1660. (35) Fehr, M.; Frommer, W. B.; Lalonde, S. Visualization of maltose uptake in living yeast cells by fluorescent nanosensors. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 9846.

Figure 5. Maltose addition alters the SERS spectrum of MBPAg4. (a) The SERS spectra of MBP-Ag4 (5.75 µM) were collected before (red) and after (black) addition of 100 µM maltose. Peaks exhibiting decreased intensity following maltose addition are indicated by dashed lines and labeled. Similar spectra were obtained in replicate experiments. (b) SERS spectrum of pure maltose at a 2.5 mM concentration. Spectra were collected at 100 mW power and at 514 nm with an integration time of 120 s.

increase in crystal resonant frequency over time but does not cause “wholesale” release of MBP-Ag4. Thus, MBP-Ag4 stably interacts with silver whether or not maltose is present in solution and protein desorption should be minimal over the 2 min required for SERS data collection. Figure 5 shows the MBP-Ag4 SERS spectra recorded in the presence or absence of 100 µM maltose. Although the spectra were similar, the intensity of several peaks decreased significantly in the maltose-treated sample (Figure 5a, dashed lines). Table 1 shows that most of these peaks correspond to the vibrational signatures of F, Y, and W, a result consistent with the idea that aromatic amino acids from MBP contribute to the SERS signal in addition to the single Y and F residues of the Ag4 extension. A control experiment conducted with maltose alone showed that, at high concentration, this analyte adsorbs to colloidal silver and can be detected by SERS (Figure 5b). However, none of the

Binding and Orientation of Proteins on Nanoparticles

Langmuir, Vol. 24, No. 5, 2008 2007

Figure 7. Proposed orientation of MBP-Ag4 on silver nanoparticles. The amino (N) and carboxyl (C) termini are shown in black and red, respectively.

Figure 6. Location of surface-exposed aromatic residues in the maltose-free (left) and maltose-bound (right) forms of maltosebinding protein. Views of the bottom (a), back (b), and front (c) faces of the protein are shown. The back and front views are rotated by (90° relative to the bottom view and by 180° with respect to each other. Y, F, and W residues are shown in yellow, green, and blue, respectively. Maltose is shown in orange in the liganded structures. The C-terminus of the protein to which the Ag4 extension is fused is shown in red.

maltose-characteristic peaks overlap with the spectral features that are altered when maltose is added to MBP-Ag4 (Figure 5). Furthermore, maltose does not alter the UV/visible spectrum or cause aggregation of colloidal silver (data not shown). Taken together, the above results indicate that the structural rearrangements induced by maltose binding increase the distance of a subset of MBP residues from the silver surface. Orientation of Silver-Bound MBP-Ag4. To gain insights on how MBP-Ag4 interacts with silver, the molecular surface of wild type maltose-binding protein was calculated using the coordinates of the ligand-free27 and maltose-bound28 crystal structures. Solvent-accessible F (green), Y (yellow), and W (blue) residues were mapped onto the two versions of the protein surface (Figure 6). In these images, the C-terminal amino acid of authentic maltose-binding protein (L370 shown in red) corresponds to the last S of the SSSGGG linker preceding the Ag4 sequence in the MBP-Ag4 variant (see Figure 1a). Thus, MBP-Ag4 contains an additional 15 amino acids that are not shown on the models since their structure is unknown. We estimate that the dimension of the Ag4 silver-binding extension is 3-5 nm. Considering the location of the maltose-binding protein C-terminus with respect to the overall molecule, there are three possible orientations for MBP-Ag4 immobilized onto silver: (i) standing onto its “bottom” face with “front”, “back”, and “top” faces exposed to solvent (using the convention of Figure 6); (ii) lying on its back face and exposing its front face to solvent; or (iii) lying on its active site-containing front face with a solventexposed back face. Figure 6a shows that the degree of exposure of F, Y, and W residues on the bottom face is comparable in the two structures. On the back face (Figure 6b), conformational

changes induced by maltose binding cause an increase in the solvent accessibility of a number of aromatic residues (in particular F92, F169, Y99, Y167, Y171, Y176, Y242, W94, W158). On the other hand, there is a significant decrease in the exposure of Y (Y70, Y99, Y155, Y341) and W residues (W62, W230, W340) on the front face upon maltose binding. Because only the latter is consistent with SERS data, we propose that the active sitebearing, “front” face of MBP-Ag4 is oriented toward the silver surface, as depicted in Figure 7. Although further work will be needed to determine if this is the case, Figure 7 also suggests that the Ag4 extension may be running along the central cleft defined by the two structural lobes of the maltose-binding protein.

Conclusion Compared to other analytical methods that have been exploited to explore the conformation of proteins at interfaces (e.g., AFM, solid-state NMR, and ToF-SIMS),36 SERS is a simple, rapid and sensitive technique that can be used in physiologically relevant aqueous environments and does not require protein labeling. Unfortunately, SERS studies of large proteins at interfaces have been complicated by their differential affinity and random adsorption on the enhancing surface. We have shown here that fusion of a silver-binding peptide to the C-terminus of maltosebinding protein improves the quality of SERS data by increasing the adhesion and controlling the orientation of MBP on colloidal silver. Considering that the amino and carboxyl termini of most proteins tolerate short amino acid extensions and that internal permissive sites are not uncommon,19,20 such a genetic engineering approach may prove useful to investigate the conformation of other polypeptides at SERS-active surfaces and to specify protein orientation for nanobiotechnology applications. It is also worth pointing out that, because phage and cell surface display can be used to identify peptides exhibiting high affinity for virtually any inorganic materials,37-39 the genetic route opens the door to the use of less traditional SERS enhancers (e.g., copper, aluminum, platinum, and rhodium). Together with progress in controlling particle shape through innovative chemistry and nanofabrication,7,8 the possibility of using materials of different compositions should facilitate the development of enhancers exhibiting highly tailored plasmon resonance bands for sensing and diagnostics applications. (36) Gray, J. J. The interaction of proteins with solid surfaces. Curr. Opin. Struct. Biol. 2004, 14, 110. (37) Baneyx, F.; Schwartz, D. T. Selection and analysis of solid-binding peptides. Curr. Opin. Biotechnol. 2007, 18, 312. (38) Sarikaya, M.; Tamerler, C.; Jen, A. K.; Schulten, K.; Baneyx, F. Molecular biomimetics: nanotechnology through biology. Nat. Mater. 2003, 2, 577. (39) Sarikaya, M.; Tamerler, C.; Schwartz, D. T.; Baneyx, F. Materials assembly and formation using engineered polypeptides. Annu. ReV. Mater. Res. 2004, 34, 373.

2008 Langmuir, Vol. 24, No. 5, 2008

Acknowledgment. We are grateful to Pete Laxton for help with photon correlation spectroscopy experiments and acknowledge the help of the University of Washington Nanotech User Facility, a member of the National Science Foundation National Nanotechnology Infrastructure Newtork. This work was supported

Sengupta et al.

by the University of Washington Genetically Engineered Materials Science and Engineering Center (GEMSEC), a National Science Foundation MRSEC. LA702079E