Multicolor Surface Plasmon Resonance Imaging of Ink Jet-Printed

Jun 15, 2007 - Chung-Tien Li, Kun-Chi Lo, Hsin-Yun Chang, Hsieh-Ting Wu, Jennifer H. Ho, Ta-Jen Yen. Ag/Au bi-metallic film based color surface plasmo...
0 downloads 0 Views 405KB Size
Download PDF
Anal. Chem. 2007, 79, 5124-5132

Multicolor Surface Plasmon Resonance Imaging of Ink Jet-Printed Protein Microarrays Bipin K. Singh and Andrew C. Hillier*

Department of Chemical and Biological Engineering and Department of Chemistry, Iowa State University, Ames, Iowa 50011

We report a technique that utilizes surface plasmon resonance dispersion as a mechanism to provide multicolor contrast for imaging thin molecular films. Illumination of gold surfaces with p-polarized white light in the Kretschmann configuration produces distinct reflected colors due to excitation of surface plasmons and the resulting absorption of specific wavelengths from the source light. In addition, these colors transform in response to the formation of thin molecular films. This process represents a simple detection method for distinguishing between films of varying thickness in sensor applications. As an example, we interrogated a protein microarray formed by a commercial drop-on-demand chemical ink jet printer. Submonolayer films of a test protein (bovine serum albumin) were readily detected by this method. Analysis of the dispersion relations and absorbance sensitivities illustrate the performance and characteristics of this system. Higher detection sensitivity was achieved at angles where red wavelengths coupled to surface plasmons. However, improved contrast and spatial resolution occurred when the angle of incidence was such that shorter wavelengths coupled to the surface plasmons. Simplified optics combined with the robust microarray printing platform are used to demonstrate the applicability of this technique as a rapid and versatile, high-throughput tool for label-free detection of adsorbed films and macromolecules.

Surface plasmon resonance (SPR) sensors have become increasingly popular as a label-free detection platform for applications in immunosensing,1-3 proteomics,4,5 drug discovery,6,7 and detection of environmental pollutants.8,9 SPR sensing has also been * To whom correspondence should be addressed. E-mail: hillier@ iastate.edu. (1) Nishimura, T.; Hifumi, E.; Fujii, T.; Niimi, Y.; Egashira, N.; Shimizu, K.; Uda, T. Electrochemistry 2000, 68, 916-919. (2) Hsieh, H. V.; Stewart, B.; Hauer, P.; Haaland, P.; Campbell, R. Vaccine 1998, 16, 997-1003. (3) Mullett, W. M.; Lai, E. P. C.; Yeung, J. M. Methods 2000, 22, 77-91. (4) Oda, Y.; Owa, T.; Sato, T.; Boucher, B.; Daniels, S.; Yamanaka, H.; Shinohara, Y.; Yokoi, A.; Kuromitsu, J.; Nagasu, T. Anal. Chem. 2003, 75, 2159-2165. (5) Wilkinson, F. L.; Holaska, J. M.; Zhang, Z. Y.; Sharma, A.; Manilal, S.; Holt, I.; Stamm, S.; Wilson, K. L.; Morris, G. E. Eur. J. Biochem. 2003, 270, 2459-2466. (6) Rich, R. L.; Day, Y. S. N.; Morton, T. A.; Myszka, D. G. Anal. Biochem. 2001, 296, 197-207. (7) Baird, C. L.; Courtenay, E. S.; Myszka, D. G. Anal. Biochem. 2002, 310, 93-99.

5124 Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

applied in detecting biologically relevant processes such as DNA hybridization10,11 and protein-DNA interactions.12,13 Advantages of SPR-based methods include small reagent volumes, real-time analysis, and the ability to eliminate labeling reactions.14 Details of various SPR-based platforms and their applications can be found in recent reviews.15-18 Most reports using SPR detection involve single-color formats where changes in the intensity of reflected light are monitored at a fixed angle of incidence near the reflectivity minimum or reflectivity is measured as a function of incident angle.19 Twocolor SPR sensing has been developed to simultaneously determine the refractive index and thickness of adsorbed films.11,20 The capabilities of multicolor SPR experiments have not been reported in detail except for a single example where surface plasmoninduced color changes were used to image a patterned substrate in the visible spectrum.21 Multicolor SPR imaging is based upon the absorption of specific wavelengths from incident white light due to coupling to the surface plasmons (SPs).22 This adsorption changes the color of the reflected light depending on the wavelengths absorbed. A change in the refractive index near the surface of the SPR sensor, which can occur due to binding of molecules to the surface or the formation of thin films, causes a shift in the absorbed wavelengths, which, in turn, changes the color of the reflected light. Although the details of the color change are complex, the impact on the reflected light spectrum can be recognized as a simple color change, which can be easily perceived by a human operator or analyzed digitally using methods similar (8) Rodriguez-Mozaz, S.; Marco, M. P.; de Alda, M. J. L.; Barcelo, D. Anal. Bioanal. Chem. 2004, 378, 588-598. (9) Baeumner, A. J. Anal. Bioanal. Chem. 2003, 377, 434-445. (10) Nilsson, P.; Persson, B.; Uhlen, M.; Nygren, P. A. Anal. Biochem. 1995, 224, 400-408. (11) Peterlinz, K. A.; Georgiadis, R. M.; Herne, T. M.; Tarlov, M. J. J. Am. Chem. Soc. 1997, 119, 3401-3402. (12) Silin, V.; Plant, A. Trends Biotechnol. 1997, 15, 353-359. (13) Gotoh, M.; Hasebe, M.; Ohira, T.; Hasegawa, Y.; Shinohara, Y.; Sota, H.; Nakao, J.; Tosu, M. Genet. Anal.: Biomol. Eng. 1997, 14, 47-50. (14) Revoltella, R. P.; Robbio, L. L.; Liedberg, B. Biotherapy 1998, 11, 135145. (15) Green, R. J.; Frazier, R. A.; Shakesheff, K. M.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B. Biomaterials 2000, 21, 1823-1835. (16) Lechuga, L. M.; Calle, A.; Prieto, F. Quim. Anal. 2000, 19, 54-60. (17) Homola, J. Anal. Bioanal. Chem. 2003, 377, 528-539. (18) Homola, J.; Koudela, I.; Yee, S. S. Sens. Actuators, B 1999, 54, 16-24. (19) Brockman, J. M.; Nelson, B. P.; Corn, R. M. Annu. Rev. Phys. Chem. 2000, 51, 41-63. (20) Peterlinz, K. A.; Georgiadis, R. Opt. Commun. 1996, 130, 260-266. (21) Knobloch, H.; Szada, v.; Borryszkowski, G.; Woigk, S.; Helms, A.; Brehmer, L. Appl. Phys. Lett. 1996, 69, 2336-2337. (22) Raether, H. Surface plasmons on smooth and rough surfaces and on gratings; Springer-Verlag: Berlin, 1988. 10.1021/ac070755p CCC: $37.00

© 2007 American Chemical Society Published on Web 06/15/2007

to those reported for lithium ion sensing optodes23,24 and oxygen sensors.25 SPR imaging methods have been most frequently used in conjunction with a high-throughput, array-based formats. Multichannel26 and two-dimensional array-based19,27,28 sensing platforms allow for simultaneous interrogation of large numbers of samples and can include integrated reference surfaces on the sensor. These array-type sensors have been most widely applied to perform highthroughput sensing of protein-surface interactions.29-32 SPR imaging of protein microarrays33 is now being used to decipher complex interactions of proteins with DNA34,35 and other proteins.27,36 A high degree of correlation has been found between traditional biochemistry methods such as ELISA and SPR imagingbased methodologies.36,37 In this work, we describe the details of multicolor SPR imaging for interrogation of protein microarrays. A custom-built microarray of bovine serum albumin (BSA) was prepared by chemical ink jet printing of protein solutions onto the surface of a gold-coated prism. Multicolor SPR imaging was used to detect the quantity of adsorbed protein as a function of the concentration of protein solution and number of printed droplets. SPR images of the microarray depict changes in color at a fixed angle of incidence and provide clear discrimination between the unmodified gold background and BSA-film covered regions of varying thickness. Analysis of the dispersion relations and absorbance sensitivities illustrate the performance and characteristics of this system. In particular, the impact of imaging angle and wavelength on sensitivity and spatial resolution is characterized and compared to SPR dispersion curves. EXPERIMENTAL SECTION Materials and Reagents. Bovine serum albumin (SigmaAldrich, St. Louis, MO), nitric acid (J.T. Baker, Phillipsburg, NJ), (3-mercaptopropyl)trimethoxysilane (MPTS; Aldrich, Milwaukee, WI), gold (99.999%, Ernest Fullam, Latham, NY), hydrogen peroxide, sulfuric acid, and toluene (Fisher Chemical Co., Fair Lawn, NJ) were used as received. All solutions were prepared with 18 MΩ deionized water (NANOPure, Barnstead, Dubuque, IA). (23) Hirayama, E.; Sugiyama, T.; Hisamoto, H.; Suzuki, K. Anal. Chem. 2000, 72, 465-474. (24) Suzuki, K.; Hirayama, E.; Sugiyama, T.; Yasuda, K.; Okabe, H.; Citterio, D. Anal. Chem. 2002, 74, 5766-5773. (25) Evans, R. C.; Douglas, P. Anal. Chem. 2006, 78, 5645-5652. (26) Berger, C. E. H.; Beumer, T. A. M.; Kooyman, R. P. H.; Greve, J. Anal. Chem. 1998, 70, 703-706. (27) Kanda, V.; Kariuki, J. K.; Harrison, D. J.; McDermott, M. T. Anal. Chem. 2004, 76, 7257-7262. (28) Singh, B. K.; Hillier, A. C. Anal. Chem. 2006, 78, 2009-2018. (29) Schena, M. Protein microarrays; Jones and Bartlett: Sudbury, Ma, 2005. (30) Robinson, W. H.; DiGennaro, C.; Hueber, W.; Haab, B. B.; Kamachi, M.; Dean, E. J.; Fournel, S.; Fong, D.; Genovese, M. C.; de Vegvar, H. E. N.; Skriner, K.; Hirschberg, D. L.; Morris, R. I.; Muller, S.; Pruijn, G. J.; van Venrooij, W. J.; Smolen, J. S.; Brown, P. O.; Steinman, L.; Utz, P. J. Nat. Med. 2002, 8, 295-301. (31) Angenendt, P. Drug Discovery Today 2005, 10, 503-511. (32) Kingsmore, S. F. Nat. Rev. Drug Discovery 2006, 5, 310-320. (33) Kambhampati, D. Protein microarray technology; Wiley-VCH: Weinheim, 2004. (34) Brockman, J. M.; Frutos, A. G.; Corn, R. M. J. Am. Chem. Soc. 1999, 121, 8044-8051. (35) Shumaker-Parry, J. S.; Aebersold, R.; Campbell, C. T. Anal. Chem. 2004, 76, 2071-2082. (36) Cherif, B.; Roget, A.; Villiers, C. L.; Calemczuk, R.; Leroy, V.; Marche, P. N.; Livache, T.; Villiers, M. B. Clin. Chem. 2006, 52, 255-262.

SPR Sensor Chip Preparation. The 1 × 3 in. glass slides (Fisher Scientific) were cut to size, cleaned in 2% detergent solution (Neutrad, Decon Laboratories), and then immersed in piranha solution (H2SO4/H2O2 30%, 75:25 v/v, Caution: piranha solution reacts violently with organic compounds and must be handled with extreme care) at 50 °C for 30 min. This was followed by copious rinsing with deionized water and drying under a stream of nitrogen. An adhesion layer was then formed by incubating the cleaned glass slides in a 5 mM solution of MPTS in toluene for 6 h. The slides were rinsed in toluene to remove unbound MPTS and dried in nitrogen. The modified slides were subsequently placed in a vacuum chamber for deposition of ∼50 nm of gold by resistive heating (Model Bench Top Turbo III, Denton Vacuum, Moorestown, NJ) at a rate of 0.1 Å s-1. The gold-coated SPR chips were subsequently cleaned by exposing their surfaces to an oxygen plasma for 2 min prior to use. This plasma treatment served to remove surface contaminants and create a reproducibly hydrophilic surface. Protein Microarray Construction. A solution of BSA in water was prepared at a concentration of 0.25 mg mL-1. A BSA solution with a concentration of 0.025 mg mL-1 was prepared by making 1:10 dilution of the 0.25 mg mL-1 BSA solution. The solutions were filtered through 0.2-µm syringe filters (Varian, Inc., Palo Alto, CA) prior to use to avoid clogging of the microjet tips. The protein microarray used in this study was printed using a chemical ink jet printer that utilizes the principle of piezoelectric dispensing (JetlabII, Microfab Technologies, Inc., Plano, TX).38 Microarrays were constructed by printing individual drops at predetermined locations on the SPR sensor chip in a 6 × 10 array format. After rapid evaporation of water from the dispensed drops, a thin film of BSA was formed on the SPR sensor chip. The top three rows of the array were printed using 0.25 mg mL-1 BSA solution while the bottom three rows were printed using 0.025 mg mL-1 BSA solution. The first spot in each row was formed by printing a single drop, and the number of drops dispensed at each successive spot was increased by one. Thus, the last (tenth) spot in each row was made by printing 10 drops at its location. Since the dispensed drops dried in ∼3 s, a 10-s time gap was used between consecutive drops at the same location to ensure complete drying of the previously printed spot. The row and column spacing of the spots was set as 200 µm. The same nozzle was used for all printing to avoid any variation in the size and shape of drops due to changes in the microjet device. Estimation of Camera Response. The sensitivities of red, green, and blue light detection elements of the camera (Canon EOS Digital Rebel XT) used in SPR imaging were estimated using a prism (see Supporting Information). A collimated beam from a tungsten halogen light source was passed through a slit and the resulting sheet of light directed incident to one face of a BK7 equilateral prism (Edmund Optics, Barrington, NJ). The white light was dispersed to create a spectrum of constituent wavelengths, which was subsequently imaged by the camera. Neutral density filters (Newport) were employed to avoid saturation of (37) Unfricht, D. W.; Colpitts, S. L.; Fernandez, S. M.; Lynes, M. A. Proteomics 2005, 5, 4432-4442. (38) Sloane, A. J.; Duff, J. L.; Wilson, N. L.; Gandhi, P. S.; Hill, C. J.; Hopwood, F. G.; Smith, P. E.; Thomas, M. L.; Cole, R. A.; Packer, N. H.; Breen, E. J.; Cooley, P. W.; Wallace, D. B.; Williams, K. L.; Gooley, A. A. Mol. Cell. Proteomics 2002, 1, 490-499.

Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

5125

the sensor. The positions of select wavelengths in the spectrum were identified with a set of band-pass interference filters to yield a calibration of red, green, and blue pixel values with the wavelength of light being imaged. The linear response of the camera was also confirmed using a selection of neutral density filters. Imaging of Surface Plasmon Dispersion. A custom-built optical system was used to image the surface plasmon resonance response with respect to both wavelength and angle of incidence. A tungsten halogen lamp (LS-1, Ocean Optics, Dunedin, FL) was used as the white light source (360-2500 nm). The light from the lamp was collimated with a biconvex lens (focal length 100 mm) (Newport Corp., Irvine, CA), and a rotating linear polarizer (Edmund Optics, Barrington, NJ) was placed next to the lens to allow for switching between p- and s-polarization states. A custommachined slit with dimensions of 25 mm × 0.5 mm was placed in the light path to shape the light beam as a 0.5-mm thin collimated sheet. The sheet of light was made convergent by passing through a planoconvex lens (focal length 75 mm) to illuminate a BK7 hemicylindrical prism coupler (focal length 25 mm). A glass slide coated with 50-nm gold film was kept in optical contact with the flat side of the prism coupler using an index-matching liquid (Index Matching Liquid 150, Norland Products Inc., Cranbury, NJ). The reflected diverging sheet of light was recollimated using a second planoconvex lens and then passed through a transmission diffraction grating (Edmund Optics, Barrington, NJ) with the grooves aligned parallel to the sheet of light. The sheet of light was diffracted while passing through the grating, and the firstorder dispersion was captured by a digital camera equipped with a high-sensitivity and high-resolution CMOS sensor (Canon EOS Digital Rebel XT). Identification of spatial location of various wavelengths was done using a series of band-pass interference filters (Edmund Optics, Barrington, NJ) with wavelengths centered at 488, 532, 632, and 671 nm. Surface Plasmon Resonance Imaging. A custom-built device was used to collect SPR images. A 90° BK7 prism (Edmund Optics) was mounted on a rotating tilt stage for alignment and rotation to the desired angle of incidence. Manual rotation of a linear polarizer allowed switching of polarization states of the incident light between p and s. The hypotenuse face of the prism was brought into optical contact with the SPR sensor chip using a thin film of index-matching liquid. A collimated beam of white light illuminated the back side of the sample. The reflected image was magnified with a 10× objective (Mitutoyo) and imaged by a digital camera equipped with a high-sensitivity and high-resolution CMOS sensor (Canon EOS Digital Rebel XT). A 632-nm bandpass interference filter was placed in the light path for monochromatic imaging. Theoretical SPR Response. N-Phase Fresnel calculations were performed to model SPR reflectivity curves as a function of wavelength using a published procedure.39 Scripts were written in MATLAB (The Mathworks, Inc., NJ) to calculate normalized reflectivities as a function of angle of incidence and wavelength of light for three-layer (BK7/gold/air) or four-layer (BK7/gold/ film/air) models. Wavelength-dependent refractive indices of BK7 glass and gold40 were used in the calculations. The refractive index (39) Hansen, W. N. J. Opt. Soc. Am. 1968, 58, 380-&. (40) Innes, R. A.; Sambles, J. R. J. Phys. D: Met. Phys. 1987, 17, 277-287.

5126

Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

for the film was assumed to be constant at a value of 1.45. The dispersion of BK7 glass was estimated using the Sellmeier equation:41

n2 (λ) ) 1 +

B1λ2 λ2 - C1

+

B2λ2 λ2 - C 2

+

B3λ2 λ2 - C 3

(1)

where n is the refractive index of BK7, λ is the wavelength of light in µm, and the values of the constants were taken from Schott glass: B1 ) 1.039 612 12, B2 ) 2.317 923 44 × 10-1, B3 ) 1.010 469 45, C1 ) 6.000 698 67 × 10-3 µm2, C2 ) 2.001 791 44 × 10-2 µm2, and C3 ) 1.035 606 53 × 102 µm2. RESULTS AND DISCUSSION SPR sensing was performed with prism coupling in the Kretschmann configuration due to the superior sensitivity of this method among the various SPR excitation techniques.18 In this configuration, light travels through a prism to excite surface plasmons in a thin metal film from the back side of a test sample under conditions of attenuated total internal reflection. Our sample comprises a sensor chip containing a gold-coated glass slide, which was brought into optical contact with the prism using an index matching fluid. Optical excitation of SPs at the metal-air interface requires that the momentum of the incident light matches that of the SPs. When light illuminates a prism-slide assembly under the conditions of total internal reflection, the momentum of light is increased by a factor of 01/2 and can excite surface plasmons when the following equation is satisfied:42

k′sp )

x

ω c

′12 ω )  sin θ0 ) kx ′1 + 2 x 0 c

(2)

where kx is the horizontal component of light wave vector in the plane of incidence, k′sp is the real part of the surface plasmon wave vector, ω and c are angular frequency and velocity of light in vacuum, θ0 is angle of incidence, 0, 1( ) 1 + i′′1), and 2 are the dielectric constant of prism, gold, and air. Optical excitation of SPs appears as a decrease in the reflected light intensity and is a function of both the wavelength and angle of the incident light. Using a prism-coupler method ensures that a large shift in absorbed wavelength occurs for a given change in refractive index at the surface of the SPR sensor.18 White light imaging was used to directly map surface plasmon resonance dispersion on gold slides as a function of angle of incidence and wavelength of light using a custom-built optical train.43 Figure 1 illustrates the optical system used in imaging SPs (see Experimental Section). Briefly, a thin sheet of p-polarized white light was shaped into a wedge with a lens to provide a range of incidence angles at the sample. The incident light passed through a hemicylindrical prism coupler and reflected from the gold film sample. The reflected wedge of light was then shaped back into a thin sheet of light with a second hemicylindrical lens. (41) Efimov, A. M. Optical constants of inorganic glasses; CRC Press: Boca Raton, FL, 1995. (42) Raether, H. Excitation of plasmons and interband transitions by electrons; Springer-Verlag: Berlin, 1979. (43) Kitson, S. C.; Barnes, W. L.; Bradberry, G. W.; Sambles, J. R. J. Appl. Phys. 1996, 79, 7383-7385.

Figure 1. Plan view and profile view of the experimental setup used for imaging the surface plasmon dispersion of BK7/gold/air system.

The emerging light was then directed through a transmission diffraction grating to separate its constituent wavelengths in an orthogonal direction. The resulting image provided a range of incident angles along one axis and a range of wavelength values along the other. This image was focused onto a CCD camera for observation and analysis (Figure 2A). The image depicted in Figure 2A is a photograph recorded by the CCD camera of the experimentally measured angle and wavelength-dependent dispersion curve for this gold sample. The observed colors in the horizontal direction are consistent with the wavelengths given by their spatial position. The dark region passing across the image is the surface plasmon resonance response as a function of wavelength and angle of incidence. At long wavelengths, the resonance is sharp and approaches low angles of incidence. As wavelength decreases, the SPR resonance broadens and transitions to larger incident angles. SPs cannot be generated at wavelengths below ∼530 nm. To compare this measured SPR response to theoretical predictions, a reflectivity calculation based upon a wavelength-dependent prism/gold/air interface was performed.39 The calculated reflectivity was multiplied by the measured camera sensitivity (see Supporting Information) to yield Figure 2B, which is the predicted response for the dispersion image. The shape and details of the experimental result in Figure 2A and the theoretical prediction in Figure 2B exhibit a high degree of similarity. In order to compare the color SPR response to that produced by a typical monochromatic light source, reflectivity images were taken with the diffraction grating removed. The resulting images (Figure 3) depict reflectivity versus angle for s- and p-polarized light for both color and monochromatic light. The monochromatic response was acquired by placing a 632-nm interference filter in the optical path. Figure 3A and B shows the expected featureless images associated with s-polarized white light and 632-nm light. In contrast, a multicolored image was obtained with p-polarized white light (Figure 3C), and an image containing a dark band

Figure 2. (A) Image of surface plasmon dispersion captured by the digital camera. (B) Theoretical surface plasmon dispersion using calculated reflectivity values and experimentally measured calibration curves for camera.

was obtained with p-polarized 632-nm light (Figure 3D). The color image can be explained by considering that reflectivity is both angle and wavelength dependent. Line profiles of the blue, green, and red pixels of the imaging camera can be used to illustrate this behavior (Figure 3E). A yellowish-white hue corresponding to the natural reflectance of gold appears for angles of incidence less than 43°. For angles between 43° and 44.5°, the color sharply changes to green due to increased absorbance of red light. The hues gradually change to magenta for angles greater than 44.5°. The yellow-green color transition is marked by increased absorption of red colors while the green-magenta color transition is marked by gradual recovery of red values and a drop in green values. Note that the blue channel is relatively featureless, which is consistent with the lack of resonance at shorter wavelengths. This somewhat complicated color response contrasts to the Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

5127

a maximum in contrast for a given shift in the SPR response. Typically, the angle of incidence for imaging is chosen to be slightly lower than the angle of resonance minimum. Adsorption produces a positive shift in the resonance minimum. Consequently, there is a large increase in reflectivity at the chosen imaging angle.19 However, for multicolor SPR, the images represent the product of camera sensitivity and the SPR response. The overall response needs to be optimized in order to get maximum contrast in the SPR images. For multicolor SPR imaging using a three-chip CCD camera, the optimum imaging conditions can be defined as the angle providing a maximum change in one or more of the imaging colors. As noted earlier, the blue channel of our RGB camera displayed minimal response to SPR resonance. Therefore, we chose to focus on the red and green channels. Optimum image contrast can be defined by a condition where the magnitude of the changes in the red and green channels is maximized. This is achieved by looking for angles of incidence corresponding to extremes of the function

f ) (R2 - R1) - (G2 - G1)

Figure 3. (A) s-Image using converging beam of white light and (B) 633-nm light. (C) p-Image using converging beam of white light and (D) 633-nm light. (E) Intensity profile of (C) showing the response of red, green, and blue detection elements of the camera at the conditions of surface plasmon resonance. (F) Reflectivity profile of the (p/s) image calculated using (B) and (D).

characteristic resonance dip observed from a monochromatic source (Figure 3F). Notably, this sharp SPR curve demonstrates that narrow bandwidth light sources can be used in place of a strictly monochromatic light source (e.g., lasers) without any degradation in the quality of the SPR data.44 In order to determine the optimum conditions for multicolor SPR imaging, a sensitivity analysis was performed. In singlewavelength SPR, the optimal imaging angle is that which produces 5128

Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

(3)

where R1, G1 and R2, G2 are the red and green intensity values for the images captured before and after film formation. Notably, the definition of function f (eq 3) is somewhat arbitrary. Other choices could also be made that also provide a maximum in the detector signal. To compute the optimal imaging conditions, we calculated the theoretical SPR response before and after formation of 5-nm film and multiplied it with the sensitivity of the camera to emulate the camera response. Figure 4A shows the calculated function f as a function of wavelength and angle of incidence. The three solid vertical lines drawn in Figure 4A represent the wavelengths giving maximum sensitivities for the blue, green, and red channels of our CCD camera at ∼ 460, 560, and 630 nm, respectively. As can be seen in the color image, a local maximum and minimum in the function f exists along the 630-nm line between 43° and 45°. The origin of these extrema can be elucidated by considering changes with respect to angle at a fixed wavelength due to film formation. For example, Figure 4B shows the value of f versus angle at 630 nm, which is near the maximum sensitivity of the red channel of our CCD camera. A maximum in f occurs of 43.5°, which is slightly less than the angle at which SPR minimum occurs on bare gold (43.8°). This is usually chosen as the optimum angle for monochromatic imaging with a 632-nm source because the change in reflected intensity before and after film formation is the greatest.19 This is due to a large increase in reflected intensity that would be measured at the red channel. A minimum also occurs in the 630-nm SPR curve at ∼44.3°. This decrease in intensity is observed after film formation since this angle corresponds to the rising portion of SPR curve and, thus, a decreasing reflected intensity near 630 nm. The second line in Figure 4B shows the value of the function f at a wavelength (560 nm) near the maximum in sensitivity of the green channel of our CCD camera. This line profile goes through a minimum in the function f near ∼45° due to an increasing intensity in the green channel (44) Melendez, J.; Carr, R.; Bartholomew, D.; Taneja, H.; Yee, S.; Jung, C.; Furlong, C. Sens. Actuators, B 1997, 39, 375-379.

Figure 5. (A) Image of the glass capillary microjet device dispensing aqueous drops containing BSA at a concentration of 0.25 mg mL-1. The drops were constantly being dispensed at 500 Hz but appear stationary in the image due to a temporal aliasing method (stroboscopic effect). (B) Diagram showing the location and number of drops dispensed on the gold slide by the microjet device using 0.025 and 0.25 mg mL-1 BSA for the construction of the protein microarray.

Figure 4. (A) Image of calculated function f ) (R2 - R1) - (G2 G1), where R1, G1 and R2, G2 are the red and green pixel values for images before and after formation of a 5-nm-thick film on gold. The solid vertical lines represent the maximum sensitivities of the blue (B) green (G), and red (R) sensors of our CCD camera. The dashed horizontal lines correspond to angles chosen for SPR imaging. (B) Pixel values versus angle calculated for the function f at wavelengths of 632 and 559 nm.

and decreasing intensity at the red channel. These extrema in the function f represent potential conditions to maximum contrast in color SPR imaging. Two imaging angles chosen from these results were 43.8° and 45° (dotted horizontal lines in Figure 4A). The former gives a large increase in reflected intensity for the red channel and a slight increase in intensity at the green channel after film formation. The latter gives a slight decrease in the red channel and a large increase in the green channel following film formation. In each case, these represent extrema in the function f. In order to test the efficacy of the multicolor SPR imaging method, a microarray was created out of BSA. A commercial ink jet printer was used to print the protein microarray. Ink jet dropon-demand printers have proved to be extremely versatile technology to deliver extremely minute quantities of material to a surface. Indeed, ink jet delivery of viable cells,45 proteins,46 and many other materials47 has already been reported. This printing method allowed us to create well-defined spatial arrays as well as precisely

control the amount of material dispensed at a specific location. The printing nozzle of the chemical ink jet printer that we used consisted of a glass capillary fused to a piezoelectric crystal. A droplet of ink was dispensed from the nozzle when a voltage pulse was applied to the crystal. Figure 5A shows an image of the device ejecting an aqueous droplet containing 0.025% BSA by weight at a rate of 500 Hz. The diameter of the droplet is 48 ( 2 µm, which is very close to the inside diameter of the glass capillary (50 µm). This diameter corresponds to a drop volume of 58 ( 7 pL with 1.45 ( 0.175 pg of BSA contained within it. Larger or smaller droplets can be generated by using microjet devices with differing diameters. We used two solutions of BSA in water to print the protein micrarray: 0.25 and 0.025 mg mL-1. A schematic of the printing methodology is shown in Figure 5B, which shows a protein microarray with 6 × 10 spots. The top three rows were printed using a 0.25 mg/mL solution while the bottom three rows were printed using a 0.025 mg/mL solution. The first column in the sample was printed using a single drop of solution, and the number of printed drops was increased by one for each subsequent column. This resulted in a microarrray in which precise multiples of 14.5 pg (top three rows) or 1.45 pg (bottom three rows) of BSA were deposited. After printing of BSA micrarray on a gold-coated slide, it was imaged at 43.8° and 45° with p- and s-polarized light through a (45) Xu, T.; Jin, J.; Gregory, C.; Hickman, J. J.; Boland, T. Biomaterials 2005, 26, 93-99. (46) Pardo, L.; Wilson, W. C.; Boland, T. J. Langmuir 2003, 19, 1462-1466. (47) Calvert, P. Chem. Mater. 2001, 13, 3299-3305.

Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

5129

Table 1. Calculated Propagation Lengths and Skin Depths Associated with Surface Plasmon Resonance at Air-Gold Interface wavelength (nm)

propagation length (µm)

skin depth (nm)

559 632 700

1.6 10.0 25.2

212 328 439

toward higher angles. Thus, film-covered regions will appear yellow or the color associated with lower wavelengths at that particular angle of incidence. Another multicolor SPR image was captured at an angle of incidence of 45° which is shown in Figure 6C. In this image, green spots with a yellow rim appear on a magenta background. The spots and background colors can be explained in a similar manner. Magnified images of the protein spots are also shown in Figure 6 to illustrate the spatial resolution of SPR imaging. Each pixel in the image corresponds to ∼3 µm and is solely limited by the optics and the size of the camera sensor chip. The spatial resolution of an SPR imaging method is directly related to the propagation length Lx of the SPs, which is defined as the lateral distance in the direction of propagation at which the intensity of the SPs is reduced by (1/e) of its value at the point of excitation. The propogation length can be described by the following relationship:22

Lx )

Figure 6. SPR images of protein microarray: (A) 632-nm light at an angle of incidence of 43.8°, (B) white light at 43.8°, and (C) white light at 45°. Magnified regions of the microarray show the morphology of the protein spots created by printing nine successive drops of 0.025 and 0.25 mg mL-1 BSA solutions.

prism coupler. At 43.8°, images were also captured using a 632nm band-pass filter to directly compare the sensitivities of the color SPR method with monochromatic SPR imaging. With s-polarized light, no features of the microarray were discernible. However, the p-images clearly showed the features of the protein microarray (Figure 6). All the spots of the microarray appear oval in shape due to the oblique angles of incidence used in the experiments. Figure 6A was captured at 43.8° using 632-nm light, which is very close to the SPR minimum angle. The incident light is completely absorbed in exciting SPs at bare gold regions at this angle of incidence, which results in a dark background. At locations where protein is deposited, the SPR angle is shifted away from 43.8°, which results in an increase in the reflected light intensity. Figure 6B shows the image when the 632-nm filter was removed and the microarray was illuminated by white light. Yellow spots on the green background can be clearly seen in the image. The green color of the background is a result of absorption of red wavelengths by SPR resonance on the bare gold surface. The yellow color depicts the protein spots. The observed spot colors are due to the shift in the SPR resonance curve with respect to both angle and wavelength. With film adsorption, the SPR resonance shifts to higher angles and longer wavelengths. For regions covered with film, the color palette, as indicated in Figure 3C, is translated 5130

Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

(( ) 2π ′12 λ ′1 + 2

3/2

)

-1

′′1 2

(′1)

(4)

where Lx is the propagation length of the SPs and other symbols are defined as in eq 2. In contrast to propagation length, the spatial extent of the electromagnetic field of the SPs is measured in terms of skin depth, which is defined as depth at which the magnitude of the field falls (1/e) times. A larger skin depth provides an increased dimension of the sensing volume of the SPR detection method. A list of calculated propagation lengths and associated skin depths for SPs at the gold-air interface is presented in Table 1. A monotonically increasing relationship can be established between the wavelength of source light and the propagation lengths of the SPs as well as the skin depths. For enhanced sensitivity, longer wavelengths are desirable while shorter wavelengths are preferable for spatially resolving imaging methods.48 The increased spatial resolution when lower wavelengths are coupled to the SPs can be clearly seen from the details of the magnified spots of Figure 6. Each of those spots was created by evaporation of nine successive solution drops. The protein spots formed from 0.025 mg mL-1 BSA solution appear very homogeneous in surface concentration in contrast to the spots formed from 0.25 mg mL-1 BSA solution. The spots formed with 0.25 mg mL-1 BSA solution demonstrate the well-known phenomenon of increased concentration of material at the periphery of a drying drop, which is attributed to a capillary flow of solvent to replenish the drying pinned end of the drop.49 In addition, 6-8 subspots (48) Berger, C. E. H.; Kooyman, R. P. H.; Greve, J. Rev. Sci. Instrum. 1994, 65, 2829-2836. (49) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.; Witten, T. A. Nature 1997, 389, 827-829.

Figure 7. (A) Line profile of reflectivity along the fourth row of the protein microarray in Figure 6A revealing the increase in reflectivity with increase in mass of deposited protein along the row. (B) Plot of average reflectivity as a function of surface concentration and thickness of BSA along the row.

can be noticed near the inside of the ring, and the overall deposition pattern is very similar to literature reports.50 Note that the structure of the ring as well as the nonhomogeneous structure of the smaller features within the spot are very well resolved in Figure 6C, which corresponds to the image acquired at 45°. However, the interior features are much less well-resolved in Figure 6A or B. This is due to the majority of image contrast in Figure 6A and B coming from adsorption of longer wavelength, red light, thus giving a larger propagation length for the SPs and, thus, lower spatial resolution. In addition to the enhanced spatial resolution achieved while imaging at 45°, there is an increased “dynamic range” or an expansion in the range of colors expressed by the SPR image. At any angle, the only colors expressed by an SPR image are those corresponding to lower angles. So, at 43.8°, the colors that can be expressed in an SPR image are green and yellow (Figure 3C). Thus, both the center and the border of the spot appear yellow in Figure 7B despite the presence of a greater quantity of BSA at the peripheral ring. At 45°, the SPR images become capable of expressing magenta, green, and yellow colors (Figure 3C). Thus, the center of the spot is green surrounded by a yellow ring in Figure 6C. Alternatively, the enhancement in number of colors due to SPR in a single image can be explained by Figure 3E. The magnitude of the slope of only the red profile is large near 43.8°, while both red and green profile curves are very steep at 45°. Thus, film formation will have a significant impact on both colors and increase the variety of color combinations that will appear. (50) Deegan, R. D. Phys. Rev. E 2000, 61, 475-485.

To quantitatively characterize and compare the monochromatic to multicolor SPR imaging techniques, we investigated the sensitivity and detection limits. Figure 7A shows a plot of the intensity profile along the fourth row of Figure 6A, corresponding to the monochromatic SPR image. On average, an increase in reflectivity of 4.2% is observed between each successive spot. Since each of the dispensed drops contained 1.45 pg of BSA, the average surface concentration of BSA (ΓBSA) could be determined for the spots by measuring their diameters. An upper bound on the values of film thickness and ΓBSA can be estimated by taking into account the 14 × 4 × 4 nm ellipsoidal structure of a single BSA molecule.51 Assuming that BSA has been deposited with the longest axis parallel to the surface, the surface covered by one BSA molecule is 14 × 4 nm2. This gives the value for ΓBSA as 1.97 ng mm-2. Note that actual thickness as well as ΓBSA will be slightly less than 4 nm since neither any voids nor any protein unfolding has been accounted for in the calculations. Indeed, the thickness of a BSA monolayer has generally been measured to be slightly less than 4 nm.12,26 A plot of reflectivity versus concentration of BSA on the surface is shown in Figure 7B. For comparison, a theoretical increase in reflectivity with increasing film thickness is also shown in the same plot. The plot indicates that the first nine spots in the row have a surface concentration of less than one. A straight line through the first five points with low BSA surface coverage gave a limit of detection of 15 pg mm-2 or ∼0.1-Å film. This corresponds to an effective refractive index unit (RIU) change of 3 × 10-5 RIU assuming a skin depth of 300 nm in air and using the following equation to estimate effective refractive index:52

neff )

∫

2 lskin

∞

0

n(z) exp(2z/lskin) dz

(5)

This limit of detection is equivalent to 1.35 × 10-5 RIU based on the conversion factor of 0.9 × 10-6 RIU (pg mm-2)-1.53 However, the noise floor in the acquired images restricts the actual limit of detection to 68 pg mm-2, which is equivalent to 1.36 × 10-4 RIU from method 152 or 6.12 × 10-5 RIU from method 2.53 Figures 8A and B shows the red and green intensity profiles for the fourth row of Figure 6B and C, corresponding to color images taken at 43.8° and 45°, respectively. At 43.8°, the SPR information is carried solely by the red pixel values. Note that the limit of detection of the imaging system is reduced to ∼150 pg mm-2 (method 1, 3 × 10-4 RIU; method 2, 1.35 × 10-4 RIU) by removing the band-pass filter. However, at 45°, both red and green wavelengths excite the SPs (Figure 4E). At 45°, the slope of red profile is positive in Figure 4E, which explains the inverse relationship between ΓBSA and the intensity of reflected red light. Conversely, the reflectivity of green light increases with increased BSA concentration due to a negative slope of green profile in Figure 4E. The limit of detection at this angle of incidence has decreased to ∼400 pg mm-2 (method 1, 8 × 10-4 RIU; method 2, 3.6 × 10-4 RIU) at the expense of higher spatial resolution but is still sufficient to enable simple visual detection of a fourth of a (51) Peters, T. Adv. Protein Chem. 1985, 37, 161-245. (52) Jung, L. S.; Campbell, C. T.; Chinowsky, T. M.; Mar, M. N.; Yee, S. S. Langmuir 1998, 14, 5636-5648. (53) Stenberg, E.; Persson, B.; Roos, H.; Urbaniczky, C. J. Colloid Interface Sci. 1991, 143, 513-526.

Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

5131

Figure 8. (A) Line profile of red and green pixel values along the fourth row taken from Figure 6B at 43.8° and (B) Figure 6C at 45°.

BSA monolayer. Notably, these detection limits can be improved by using a higher quality color camera with more sensitive detector elements. CONCLUSIONS We have described a method that utilizes a white light source to generate SPR and utilizes the corresponding changes in the color of the reflected light to differentiate between surfaces that are covered with films of varying thickness. The spatial resolving capabilities and the limits of detection that were demonstrated

5132

Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

by this method make it suitable for applications in SPR microscopy and in hand-held, miniaturized SPR sensors. Although the ultimate sensitivity of this method is lower than single-wavelength SPR, the reduced number of optical components required for color SPR is potentially advantageous for microsensing or inexpensive, “dipstick”-type sensors. Moreover, a colored optical readout in dipstick-type devices is much easier to detect by the human eye than subtle changes in light intensity at a single wavelength. The reported method of multicolor SPR imaging can also be used for a simultaneous determination of the refractive index and thickness of adsorbed films in a manner that is analogous to the reported two-color SPR method.11,20 The ability of this platform to couple any wavelength of light in red and green regions of spectrum to SPs also makes this technique more flexible and convenient for this application. The limit of detection of this method was determined to be as good as 150 pg/mm2 or 3 × 10-4 RIU using a custom-made microarray with spots containing measured quantities of BSA. With this limit of detection, this technique is capable of sensing less than a monolayer of proteins while providing a simple color-readout method for the interpretation of surface film formation. ACKNOWLEDGMENT The authors gratefully acknowledge Iowa State University for partial support of this work. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review April 16, 2007. Accepted June 4, 2007. AC070755P