Surface Enhanced Raman Spectroscopy Detection of Biomolecules Using EBL Fabricated Nanostructured Substrates

Collecting control Raman spectra for the main components, including free proteins in solution and free ligands in solution or in powder form without using metal-containing substrates, is important to enable a proper comparison as well as for interpretation purposes. Figure 5A presents a typical Raman spectrum for free protein A in DI water on a glass slide without nanostructured substrates. Two bands with the highest Raman intensity, the band at 2,931 cm^-1 and at 1,091 cm^-1, correspond to vibrations involving C-H and C-S bonds, respectively. Other bands with a lower Raman intensity such as 563 cm^-1, 1,450 cm^-1, 1,653 cm^-1 and 2,426 cm^-1, can be attributed to a superposition of vibration modes^18,33,34,35. Control Raman spectra for the ligand free GPB in buffer solution with three different concentrations, 0.3, 0.9 and 1.3 mM, are shown in Figure 5B. In the figure, the broad band around 3,400 cm^-1 corresponds to the solvent³⁶, whereas the band at 2,935 cm^-1represents vibrations involving C-H bonds of the protein^33,34. Figure 5C shows the high wavelength Raman spectrum for D-glucose in buffer solution for different concentrations: 1, 6, 100, 200, and 400 mM. When the concentration of glucose is increased, C-H bonds vibration bands arise at 2,890 cm^-1 and 2,960 cm^-1. Control Raman spectra of dopamine in crystal form obtained with both 532 nm and 780 nm excitation wavelengths are shown in Figure 5D. Much of the Raman spectrum comes from the benzene ring bending and C-H bond stretches of the molecule^20,37. Some of the bands at around 3,000 cm^-1 are only observed at the 532 nm but not 780 nm excitation wavelength.

Figure 5. Control Raman spectrum of protein A in DI water obtained at 532 nm excitation wavelength¹⁸ (A); Raman spectra of ligand-free glucose binding protein in buffer solution obtained at 532 nm excitation wavelength (B); Raman spectra of D-glucose in buffer solution obtained at 532 nm excitation wavelength (C); and spectra of dopamine powder obtained at 532 nm and 780 nm excitation wavelengths²⁰(D). All the spectra are regular Raman spectra of solutions (a,b,c) and powder (d) on glass slides without nanostructured substrates. In (D), the assignment of Raman shift regimes to various molecular vibrations was done using General Atomic and Molecular Electronic Structure System (GAMESS)³² and MacMolPlt³¹ software as detailed elsewhere20. Reprinted panel (a) with permission from¹⁸ American Vacuum Society. Please click here to view a larger version of this figure.

In order to obtain SERS spectra for surface-immobilized biomolecules, substrates comprising metallic nanostructures on fused silica supports are fabricated as described in steps 1-3. The quality of fabricated substrates is monitored using scanning electron microscopy (SEM). The standard SEM procedures are described elsewhere^16,17,18,19 and not included in the present protocol. Figure 6 shows representative SEM images of Au and Ag nano-dots and nano-hexagon like structures (panels a-d), as well as non-structured Au and Ag pads (panels e and f, respectively). The next steps involve immobilization of biological material on the nano-structured substrates employing three designs listed in Table 1, and acquisition of their SERS spectra. In order to maintain an aqueous environment during Raman imaging, each sample is placed in its corresponding solution (see Table 1), capped with a thin glass cover, and sealed as illustrated in Figure 7.

Figure 6. Scanning electron microscope (SEM) images of 10 nm thick Au and Ag surface nanostructures employed as SERS substrates: (A,B) arrays of nanodots; (C,D) arrays of nano-hexagons; (E,F) unstructured pads. Substrates with Au structures (left) employ FS supports, whereas substrates with Ag structures (right) use a 10 nm thick Ni coating on the FS. The images were obtained as described elsewhere^16,17,18. Please click here to view a larger version of this figure.

Figure 7. Scheme of water-proof chamber for Raman imaging of bio-functionalized samples in solution.

In Design 1, recombinant protein A is immobilized on the substrates functionalized by a self-assembled monolayer of 11-mercaptodecanoic acid (MUA) in DI water¹⁸. The substrates in this design comprise three arrays of Au dots with a 50 nm pitch and varying inter-dot distances on fused silica (see Figures 6A and 8). The process of protein immobilization starts with the formation of a SAM on the substrates. To obtain covalent binding between the SAM and the protein, the carboxylic acid groups of SAMs are transformed into amine reactive NHS-ester by treatment with a mixture of N-ethyl-N'-(3-(dimethylamino) propyl) carbodiimide (EDC) solution and N-hydroxysuccinimide (NHS) solution in DI water. Immobilization of protein A occurs by displacement of the NHS group by lysine residues of the protein³⁸. An example of imaging samples for Design 1 with protein A immobilized on Au nanostructures is shown in Figure 9. Figure 9A presents an optical microscope image of the samples, which comprise three arrays of bio-functionalized Au nanodots with different inter-dot gaps (see also Figures 3Aa and 8), and Figure 9B shows the Raman spectral mapping over these arrays. It can be seen that the highest Raman intensities are found for Array I where the inter-dot gaps are the narrowest, whereas lower intensities are obtained for Array III with the widest inter-dot gaps. This can be explained by a stronger plasmon coupling effect produced by higher electric fields in the narrow spaces between the dots¹⁸. Figure 9C shows the strongest SERS spectra obtained for Arrays I and II. The spectra exhibit several bands (1,630 cm^-1, 1,964 cm^-1, 2,280 cm^-1, 2,577 cm^-1, and 2,916 cm^-1) in proximity to the Raman modes of free protein A in solution seen in Figure 5A. Attributable to vibrations of various bonds found in proteins, these bands either appear at similar locations in both immobilized protein and in solution, or are slightly shifted to somewhat higher wavenumbers when immobilized. In contrast, SERS spectra of similar nanostructured substrates functionalized by MUA SAM without the protein show an entirely different pattern¹⁸, confirming that Figure 9 represents SERS mapping of surface-immobilized protein A. Please click here to view a larger version of this figure.

Figure 8. SEM images of three arrays of Au nanodots on FS substrate used in Design 1¹⁸. The arrays have the same 50 nm pitch and slightly different dot radii resulting in different widths of inter-dot gaps. This is achieved by applying different EBL exposure doses to produce PMMA masks for the three arrays (see Table 1). Higher exposure doses result in wider holes in PMMA masks, allowing for larger Au dot sizes after metallization and liftoff. Reprinted with permission from¹⁸ American Vacuum Society. Please click here to view a larger version of this figure.

Figure 9. SERS imaging of substrate-immobilized protein A in Design 1¹⁸. (A) Optical microscope image of the sample, comprising three arrays of Au nanodots with 50 nm pitch and different inter-dot gaps on a fused silica substrate (see also Figure 6), bio-functionalized as shown in Figure 3A. (B) Raman mapping of the sample. (C) SERS spectra from dot Array I and II. In panel (B), the vertical axis represents the distance across the substrate, the horizontal axis represents the Raman shift, and the legend bar indicates the Raman intensities. Vertical dashed lines in panels (B) and (C) represent benchmark Raman bands of free protein A in solution and (*) in panel (C) indicates SERS bands from Array I. The Raman spectra were obtained 532 nm excitation wavelength. Reprinted with permission from¹⁸ American Vacuum Society. Please click here to view a larger version of this figure.

In Design 2, recombinant glucose binding protein (GBP)²² complexed with D-glucose (ligand) is immobilized on the appropriate substrates in potassium phosphate buffer solution. Samples with immobilized ligand-free GBP are also prepared for comparison. In this design, glucose-binding protein is attached to surface by means of a histidine tag, which binds well to Ni but not to noble metals⁶. Since the substrates comprise arrays of Ag nano-dots, nano-hexagons, and unstructured Ag pads on Ni-coated FS (Figures 6B, 6D, and 6F, respectively), one can expect most of immobilized protein molecules to be located in gaps between Ag nanostructures where Ni coating is available. The Raman spectra obtained for immobilized glucose-free and glucose-bound GBP are shown in Figures 10A and 10B, respectively. All these spectra exhibit a broad band at approximately 3,300 cm^-1, which corresponds to the buffer solution³⁶. The spectra obtained with an unstructured Ag pad contain only this single band and do not show any protein vibration modes, confirming that immobilized protein is not found on the Ag surface as expected. In contrast, the spectra obtained with arrays of Ag nano-dots and nano-hexagons exhibit bands around 1,550 cm^-1 and 2,900 cm^-1, which represent the analyte^33,34. In particular, the broad band around 1,550 cm^-1, known as the amide II band, is attributable to peptide bonds vibrations in proteins^34,35. In the case considered, this band represents a superposition of the vibration modes from GBP immobilized on Ni surface between Ag features, and is indicative of SERS enhancement of these modes in the vicinity of noble metal nanostructures when the substrates containing nano-dots or nano-hexagons are used. This band is very weak for the protein in solution in the absence of SERS enhancement (Figure 5B) and absent on Ag pads without Ni surface available for the protein binding, but it is well pronounced for nanostructured substrates with some Ni surface accessible for the protein to bind. However, even more important for the present study are the other, narrower bands around approximately 2,900 cm^-1 that can be attributed to C-H bond wibrations^33,34. The spectrum of glucose-free GBP shows a pronounced band at 2,933 cm^-1 with the nano-dots substrate, and a weak but discernible band at a similar wavelength with the nano-hexagons substrate (Figure 10A). Distinct from the case of glucose free protein, the SERS spectra of glucose-bound GBP shown in Figure 10B exhibit two bands corresponding to C-H bonds vibrations regimes, at 2,850 cm^-1 and 2,910 cm^-1. The bands are well pronounced in the spectrum of glucose-bound GBP on nano-hexagons substrate, and they also can be seen in the spectrum of GBP on nano-dots substrate. The band at 2,850 cm^-1 is reasonably close to the 2,890 cm^-1 one in the control Raman spectrum from D-glucose in solution, and therefore it can be attributed to glucose bound to the protein, whereas the other band (at 2,910 cm^-1) is attributable to C-H bond vibrations of both the protein and glucose. One can conclude that difference of SERS signatures from glucose-free and glucose-bound substrate-immobilized GBP is observable in this region, and C-H bond vibrations of protein-bound glucose are detectable employing the design described.

Figure 10. SERS spectra of ligand-free (A) and ligand-bound (B) glucose-binding protein immobilized on three different substrates in Design 2. The spectra were obtained with 532 nm excitation wavelength. Please click here to view a larger version of this figure.

In Design 3, the customized dopamine binding aptamer (DBA) with thiol termination²⁴ is immobilized on the substrate in tris(hydroxymethyl)aminomethane (TRIS) ethylenediaminetetraacetic acid (EDTA) buffer solution, and the dopamine is then bound to the immobilized aptamer²⁰. The substrates for this design contain arrays of Au nano-hexagons on FS (Figure 6C). Unstructured Au pads (Figure 6E) are also used for control purposes. Since DNA is intrinsically fluorescent³⁹, 780 nm excitation wavelength is used in Design 3 to reduce this factor. In this design, the recognition element (aptamer) is not Raman active in the region of Raman shifts considered in Figure 11, whereas dopamine shows a significant Raman activity in this region. Since signal from samples exposed to only dopamine without immobilized aptamer shows no resultant dopamine bands²⁰, the observed SERS bands are expected to originate from aptamer-bound dopamine. Figure 11 compares the SERS spectra of immobilized aptamer on gold nanostructures before and after the addition of dopamine. As expected, immobilized dopamine-free aptamer does not exhibit Raman bands. In contrast, a number of pronounced Raman bands are observed for dopamine-bound immobilized aptamer. The positions of most bands in Figure 11 are close to those in crystalline dopamine, albeit with differences in amplitudes.

Figure 11. SERS spectra of ligand-free (purple line) and ligand-bound (blue line) dopamine-binding aptamer immobilized on Au nano-hexagon substrates in Design 3²⁰. The red line shows a control SERS spectrum of dopamine powder. Please click here to view a larger version of this figure.