Anal. Chem. 1998, 70, 1108-1111
Diffraction-Based Cell Detection Using a Microcontact Printed Antibody Grating Pamela M. St. John,†,‡ Robert Davis,† Nathan Cady,§ John Czajka,§,| Carl A. Batt,*,§ and Harold G. Craighead†
School of Applied and Engineering Physics, Cornell Nanofabrication Facility, and Department of Food Science, Cornell University, Ithaca, New York 14853
An optical detector has been fabricated that is specific for targeted bacterial cells, by stamping an antibody grating pattern on a silicon surface. The antibody grating alone produces insignificant optical diffraction, but upon immunocapture of cells, the optical phase change produces a diffraction pattern. This technique eliminates much of the surface modifications and the secondary immunochemical or enzyme-linked steps that are common in immunoassays. Microcontact printing provides an alternative to previously reported photolithographic-mediated antibody patterning processes and uses a photolithographic process simply to produce the elastomeric stamp. We have stamped antibodies directly onto clean native oxide silicon substrates with no other chemical surface treatments. Direct binding of the antibodies to the silicon occurs in a way that still allows them to function and selectively bind antigen. The performance of the sensor was evaluated by capturing Escherichia coli O157:H7 cells on the antibody-stamped lines and measuring the intensity of the first-order diffraction beam resulting from the attachment of cells. The diffraction intensity increases in proportion to the cell density bound on the surface. A number of different types of optical biological sensors have been reported that involve the measurement of diffraction patterns as an assay readout. Alternatively, optical detection methods can employ surface plasmon resonance,1-3 surface acoustic waves,4 and fiber optical techniques.5 An optical biological sensor significantly reduces the complexity of immunoassays by eliminat-
ing secondary fluorescent or enzymatic signal-generating systems. Using surface plasmon resonance, a diffraction grating is not required; however, the process is not amenable to batch reading of samples since a baseline reading for the system must be obtained prior to the addition of analyte. The principle of the optical diffractive detection is shown in Figure 1. The use of optical diffraction as a means to detect complex formation on a silicon wafer was applied to the detection of choriogonadotropin in serum samples.6 In this study, an antichoriogonadotropin antibody was immobilized to the silicon wafer surface and was used to capture the antigen from sample solutions. Detection of the bound choriogonadotropin was achieved by illuminating the surface with a laser that diffracts due to the formation binding of the biological material. The sensor described in this earlier paper, however, required a surface coated with an antibody that is then deactivated by ultraviolet light through a mask to generate the pattern. Reproducibility in the antibody inactivation and in the scale-up of fabrication would be problems with this approach. A number of methods have been reported for binding antibodies to silicon surfaces including the use of biotin,7,8 self-assembled monolayers,9 and chemical modification of the surface.10 Antibodies can, however, be bound to the surface of silicon without any surface modification or additional intermediary conjugates. EXPERIMENTAL SECTION Silicon Stamp. The silicon master was generated using contact photolithography and consisted of 10-µm lines and 30-µm spaces. A Si (100) wafer was used with a native oxide coating of approximately 15-20 Å. The master was generated using Shipley S1813 photoresist (2 µm thick) spun onto a silicon wafer and contact photolithography through a mask with 10-µm lines and 30-µm spaces. After exposure to 405-nm light, the wafer was developed for 1 min in Shipley MF312 diluted 1:1 in water exposed to a fluorinated trichlorosilane vapor for 30 min to passivate the exposed 10-µm lines of silicon.
* Corresponding author: (phone) 607-255-2896; (fax) 607-255-8741; (e-mail)
[email protected]. † School of Applied and Engineering Physics, Cornell Nanofabrication Facility. ‡ Present address: PE Applied Biosystems, Science and Technology, 850 Lincoln Centre Drive, Bldg. 800-1, Foster City, CA 94404. § Department of Food Science. | Present address: Qualicon, DuPont Central Research and Development, Experiment Station, P.O. Box 80357 Wilmington, DE 19880-0357. (1) Cullen, D. C.; Brown, R. G. W.; Lowe, C. R. Biosensors 1987, 3, 211-225. (2) Jonsson, U.; Fagerstam, L.; Ivarsson, B.; Johnsson, B.; Karlsson, R.; Lundh, K.; Lofas, S.; Persson, B.; Roos, H.; Ronnberg, I.; Sjolander, S.; Stenberg, E.; Stahlberg, R.; Urbaniczky, C.; Ostlin, H.; Malmqvist, M. Biotechnol. 1991, 11, 620-627. (3) Jost, J. P.; Munch, O.; Anderson, T. Nucleic Acids Res. 1991, 19, 27782780. (4) Ge, K.; Liu, D.; Chen, K.; Nie, L.; Yao, S. Anal. Biochem. 1995, 226, 207211. (5) Ferguson, J. A.; Boles, T. C.; Adams, C. P.; Walt, D. R. Nature Biotechnol. 1996, 14, 1681-1684.
(6) Tsay, Y. G.; Lin, C. I.; Lee, J.; Gustafson, E. K.; Appelqvist, R.; Magginetti, P.; Norton, R.; Teng, N.; Charlton, D. Clin. Chem. 1991, 37, 1502-1505. (7) Tiefenauer, L. X.; Kossek, S.; Padeste, C.; Thiebaud, P. Biosens. Bioelectron. 1997, 12, 213-223. (8) Koyano, T.; Saito, M.; Miyamoto, Y.; Kaifu, K.; Kato, M. Biotechnol. Prog. 1996, 12, 141-144. (9) Mooney, J. F.; Hunt, A. J.; McIntosh, J. R.; Liberko, C. A.; Walba, D. M., Rogers, C. T. Proc. Natl. Acad. Sci. U.S.A. 1996, 66, 12287-12291. (10) Williams, R. A.; Blanch, H. W. Biosens. Bioelectrons. 1994, 9, 159-167.
1108 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
S0003-2700(97)01130-X CCC: $15.00
© 1998 American Chemical Society Published on Web 02/14/1998
Figure 1. Schematic diagramming the sensor. A silicon grating master was generated using contact photolithography. An elastomer replica of the master grating was used to stamp 10-µm-wide antibody stripes separated by 30 µm onto oxidized Si. The diffraction pattern from cells attached to the antibody grating was observed by illuminating the grating ∼10° off normal with a 632.8-nm wavelength He-Ne laser.
The stamp was cast into silicon using Sylgard 184 silicone elastomer (Dow Corning). A native oxide silicon surface was treated with methylene chloride, ethanol, and finally distilled water before being stamped. Microcontact Printing. Immediately before beginning the microcontact printing procedure, a temporary increase in hydrophilicity and protein adsorption of the stamps is achieved by treatment in a low-temperature plasma cleaner/sterilizer (Harrick Scientific, PDC-32G) evacuated with a mechanical roughing pump. The rf level is set to high, and plasma is left on for 30 s. The antibody solution [0.001-0.1 mg/mL (KPL, Gaithersburg, MD) in a phosphate-buffered saline solution, pH 6.5] was swabbed or pipetted onto the stamp making sure that the stamp was wet by the liquid. Excess liquid was removed with a pipet and finally dried with a stream of nitrogen gas at 40 psi. The stamp was brought into contact with the surface with the silicon substrate for 30 min. Surface Characterization. The antibodies on the surface were localized using anti-goat fluorescein-labeled antibodies (U.S. Biochemical, Cleveland, OH) diluted 1:500. The anti-goat antibodies were allowed to incubate for 1 h and, after washing with PBS with 1% Tween, observed with a with a Nikon Labophot-2 light microscope fitted with an episcopic-fluorescence attachment EFD-3 (Nikon Corp., Tokyo, Japan). Escherichia coli O157:H7 was cultured in Luria broth and the number of colony-forming units determined by plating onto Luria agar.11 The cells were diluted in PBS (pH 7.4) with 1% Tween, and a sufficient volume of the solution (∼1 mL) was pipetted on to the silicon to cover the entire surface. The cells were left on the surface for 10-15 min and then rinsed in PBS. For direct microscopic visualization, cells were stained with acridine orange (1 mg/mL) for 10 min and after washing observed microscopically at 200×. After each step, the surface was examined under an optical microscope to check for uniformity. The silicon sample was held in a clip mounted on a x, y, z and θ stage. Diffraction measurements were performed using a 632.8nm He-Ne laser focused to 1-mm diameter on a masked area of the grating. Masked areas were ∼1.3 mm in diameter, allowing measurement of constant areas of the sample. Once the sample was illuminated with the laser, an aperture was placed in the path (11) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular cloning: a laboratory manual; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 1989.
of the diffracted light beam as to collect only the first-order signal. A silicon detector was placed directly behind the aperture so that the first-order signal was centered on the 5 mm × 5 mm active area of Newport digital power meter model 815. The detector was connected to a power meter, and the intensity of the signals were measured in microwatts. As a baseline control, measurements of the intensity from the specular, m ) 0, order were taken using a neutral density filter with an attenuation of 55 times the unfiltered beam. After diffraction measurements were taken, the number of cells within the masked regions was counted to determine the bound cell concentration per unit area. RESULTS AND DISCUSSION Direct microcontact printing is a simple means to generate a patterned, immunoreactive surface suitable for a variety of applications.12,13 An analyte bound to the patterned surface will generate a diffraction pattern if it has sufficient scattering cross section (Figure 1). After using functional derivatizing agents as a monolayer to which antibodies were bound (data not shown), experiments demonstrated that the antibody could be directly stamped on the surface, thereby eliminating chemical modifications to the silicon. A silicon surface was microcontact printed using an elastomeric stamp with a solution of antibodies in PBS. The grating consisted of a repeating pattern of 10-µm lines and 30-µm spaces, and no additional treatments were used to block (i.e., bovine serum albumin) the bare silicon spaces between the antibody-stamped lines. The period of this grating was not optimized for the wavelength used. The silicon surface is an attractive medium for antibodies to adhere and we suspect that a dipole-dipole (van der Waals) interaction exists between the antibody and the hydroxyl-terminated surface.14 Antibodies bound to the surface remain able to recognize and bind their antigen efficiently and specifically. The silicon surfaces are then washed in a saline solution after the target cells are allowed to bind to the antibody, and the complexes remain bound to the silicon. To verify that the goat anti-E. coli O157:H7 antibodies were bound only to the stamped areas, anti-goat secondary antibodies conjugated with fluorescein were allowed to bind to the surfaces. (12) Kumar, A.; Whitesides, G. M. Appl. Phys. Lett. 1993, 63, 2002-2004. (13) Kumar, A.; Whitesides, G. M. Science 1994, 263, 60-62. (14) Warkentin, P.; Walivaara, B.; Lundstrom. I.; Tengvall, P. Biomaterials 1994, 15, 786-795.
Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1109
Figure 2. Fluorescence photomicrograph of anti-E. coli O157:H7 antibody-stamped silicon surface. A silicon surface was stamped with 10-µm lines of anti-E. coli O157:H7 and anti-goat antibodies conjugated to fluorescein allowed to bind. After washing, the surface was photographed with an initial 1000× magnification (reduced 50% for publication).
Figure 3. Atomic force microscopic image of anti-E. coli O157:H7 antibody-stamped silicon surface. AFM was operated in the tapping mode and the height scale posted on the right side of the image.
Figure 2 shows a fluorescence micrograph of the silicon surface. The secondary antibodies bound specifically and only in the stamped anti-E. coli O157:H7 regions of the surface. Nonspecific binding of these antibodies to the silicon surface was reduced by the inclusion of Tween into the buffer. Intense regions of fluorescence were observed, and the definition of the stamped lines at this resolution verifies that it is possible to stamp biological molecules directly onto the surface using the microcontact printing process. Variations in the fluorescence intensity over the entire sample area are a result of an incomplete coverage. Further characterization of the antibody-stamped surface was obtained using atomic force microscopy (Figure 3). The height of the antibody-stamped regions ranged from 0 to 10 nm, and they are not uniform in coverage. The maximum height of the stamped surface is consistent with the dimensions of immunoglobulin molecules (i.e., IgG, ∼100 Å), suggesting a single albeit, incomplete monolayer was deposited. Holes in the antibody-stamped areas are evident, and the edge boundary varies. Despite these 1110 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
Figure 4. Photomicrograph of E. coli O157:H7 bound to an antibody-stamped surface. A silicon surface was stamped with 10µm lines of anti-E. coli O157:H7 antibodies. E. coli O157:H7 were allowed to bind for 30 min, and the surface was washed. The cells were stained with acridine orange and photographed at an initial 200× magnification (reduced 80% for publication).
localized variations, these samples effectively bound E. coli O157: H7 as measured by diffraction. The antibodies stamped on the surface could effectively and rapidly capture E. coli O157:H7 from solution with incubation times of less than 30 min. We used fluorescence microscopy to directly observe acridine orange nucleic acid-stained cells attached to the antibodies. Figure 4 is a photomicrograph of cells bound to an antibody-stamped surface. The cells align in rows corresponding to the grating spacing. E. coli O157:H7 bound only to regions of the oxidized silicon that had been stamped with antibodies. A direct measure of the bound E. coli O157:H7 cells on the antibody grating was obtained by measuring the diffraction intensity. A power meter was used to measure the first-order diffraction of a He-Ne laser beam from an antibody-stamped sample at different cell coverages. The diffraction pattern present with cells attached to the antibody grating could easily be seen by eye. Diffraction measurements were taken from a silicon surface stamped with anti-E. coli O157:H7 antibodies after E. coli O157:H7 cells were allowed to bind. Microscopic counts of the laser-illuminated area were also carried out. A linear relationship between the diffraction intensity and the number of bound cells is observed (Figure 5). The diffraction increased from 0.02 to 0.097 over a range of 210-470 cells/mm2. This plot was generated by taking the diffraction measurements of antibody-stamped samples to obtain a background reading for each cell concentration. When an equivalent number of Salmonella were incubated with the antibody-stamped silicon, no bound cells or any diffraction signal over background was observed (data not shown). In the present study, we have generated a diffraction grating made strictly from biological molecules. The diffraction pattern is established because antigen is bound only to the antibodystamped region. The diffraction observed when a laser is used to illuminate the micrometer-scale grating is most likely due to the phase difference between adjacent regions on the surface.
Figure 5. Diffraction intensity from increasing E. coli O157:H7 cells bound to the silicon surface stamped with anti-E. coli O157:H7 antibodies. The silicon surface was stamped with 0.1 mg/mL anti-E. coli O157:H7 antibodies and then from 104 to 107 CFU of E. coli O157: H7 was allowed to bind. The surface was illuminated with a 632-run laser and the first-order diffraction measured. The cell number on the surface was plotted as a function of diffraction intensity.
Microcontact printing is a simple means to generate micrometerscale patterns on surfaces12 and typically exploits the use of selfassembling monolayers to create a reactive surface.9,13,15 Recently more elaborate methods of directing antibodies to surfaces have (15) Mrksich, M.; Whitesides, G. M. Trends Biotechnol. 1995, 13, 228-235. (16) Delamarche, E.; Bernard, A.; Schmid, H.; Michel, B.; Biebuyck, T. Science 1997, 276, 779-781. (17) Bernard, A.; Bosshard, H. R. Eur. J. Biochem. 1995, 230, 416-423. (18) Duschl, C.: Hall, E. A. H. J. Colloid. Interface Sci. 1991, 144, 368-380. (19) Suri, C. R.; Raje, M.; Mishra, G. C. Biosens. Bioelectron. 1994, 9, 535-542.
been achieved using microfluidics.16 We have shown here that antibodies can however be directly printed onto the surface of silicon and they are probably bound through dipole interactions.17 Given the relatively nonspecific nature of this interaction it is not unexpected that there will be variability within the context of antibody coverage and functionality of the bound antibodies. Variability in the interactions between silicon and proteins has been reported18,19 as well as means to improve protein adherence to the surface using chemical modification with aminosilane.14 Although there are opportunities to improve the microcontact printing to generate optical biosensors, the current format presented in this study appears to be effective. Microcontact printing is a simple means to generate an immunoreactive microstructured surface that eliminates any prior treatment and complications due to potential interactions with a functionalized surface. The optical diffraction measurements are not susceptible to small defects, therefore, any nonuniformities resulting from the stamping process are not critical to the measurements. Although we cannot exclude any desorption of the antibodies from the surface, the interaction between the microcontact printed antibodies and the silicon surface is sufficiently robust to survive the subsequent process and yield a consistent, detectable signal. In summary, this simple sensor may have value in the detection of pathogenic bacteria including E. coli O157:H7 as well as any analyte whose mass is sufficient to generate a diffraction signal when bound on a silicon surface. Received for review October 13, 1997. Accepted January 7, 1998. AC9711302
Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1111
