Scanning Force Microscopy and Fluorescence Microscopy of

Rensselaer Polytechnic Institute, Troy, New York 12180, and Department of Biology, Union College,. Schenectady, New York 12308. ReceiVed August 16, 20...
0 downloads 0 Views 735KB Size
Download PDF
Langmuir 2006, 22, 4685-4693

4685

Scanning Force Microscopy and Fluorescence Microscopy of Microcontact Printed Antibodies and Antibody Fragments John R. LaGraff*,† and Quynh Chu-LaGraff*,‡ Department of Physics, Applied Physics, and Astronomy, Department of Chemistry and Chemical Biology, Rensselaer Polytechnic Institute, Troy, New York 12180, and Department of Biology, Union College, Schenectady, New York 12308 ReceiVed August 16, 2005. In Final Form: February 17, 2006 Unlabeled primary immunoglobulin G (IgG) antibodies and its F(ab′)2 and Fc fragments were attached to oxygenplasma-cleaned glass substrates using either microcontact printing (MCP) or physical adsorption during bath application from dilute solutions. Fluorescently labeled secondary IgGs were then bound to surface-immobilized IgG, and the relative surface coverage was determined by measuring the fluorescence intensity. Results indicated that the surface coverage of IgG increased with increasing protein solution concentration for both MCP and bath-applied IgG and that a greater concentration of IgG was transferred to a glass substrate using MCP than during physisorption during bath applications. Scanning force microscopy (SFM) showed that patterned MCP IgG monolayers were 5 nm in height, indicating that IgG molecules lie flat on the substrate. After incubation with a secondary IgG, the overall line thickness increased to around 15 nm, indicating that the secondary IgG was in a more vertical orientation with respect to the substrate. The surface roughness of these MCP patterned IgG bilayers as measured by SFM was observed to increase with increasing surface coverage. Physisorption of IgG to both unmodified patterned polydimethylsiloxane (PDMS) stamps and plasma-cleaned glass substrates was modeled by Langmuir adsorption kinetics yielding IgG binding constants of KMCP ) 1.7(2) × 107 M-1 and Kbath ) 7.8(7) × 105 M-1, respectively. MCP experiments involving primary F(ab′)2 and Fc fragments incubated in fluorescently labeled fragment-specific secondary IgGs were carried out to test for the function and orientation of IgG. Finally, possible origins of MCP stamping defects such as pits, pull outs, droplets, and reverse protein transfer are discussed.

1. Introduction Monomolecular layers of proteins patterned onto inorganic surfaces on the micro- and nanoscale can potentially serve in a number of useful capacities such as active components of biosensors,1 regulators of directed cell growth,2 and diagnostic microarrays or protein chips.3 Patterning a chip containing hundreds if not thousands of different proteins, each with its own distinctive protein-substrate interaction parameters, and having all proteins retain their unique 3D shape and biological function is a formidable challenge. Although substantial investments are being poured into the commercial development of integrating and patterning proteins on surfaces, reliability and reproducibility often remain elusive owing primarily to an incomplete understanding and control of fundamental protein-substrate interactions during patterning. Thus, one of the many challenges associated with patterning proteins onto inorganic substrates is maintaining the protein’s biological activity upon adsorption by avoiding denaturation, deformation, and inaccessible active sites.4 Immunoglobulin, IgG, has been shown to maintain significant function when adsorbed on surfaces4-10 or patterned with * E-mail: [email protected]; [email protected]. † Rensselaer Polytechnic Institute. ‡ Union College. (1) Nanofabrication and Biosystems: Integrating Materials Science, Engineering, and Biology; Hoch, H. C., Jelinski, L. W., Craighead, H. G., Eds.; Cambridge University Press: New York, 1996. Commercial Biosensors: Applications to Clinical, Bioprocess, and EnVironmental Samples; Ramsay, G., Ed.; WileyInterscience: New York, 1998. Black, J. Biological Performance of Materials: Fundamentals of Biocompatibility; Marcel Dekker: New York, 1999. (2) James, C. D.; Davis, R. C.; Meyer, M.; Turner, A.; Turner, S.; Withers, G.; Kam, L.; Banker, G.; Craighead, H. G.; Isaacson, M.; Turner, J. N.; Shain, W. IEEE Trans. Biomed. Eng. 2000, 47, 17. (3) Houseman, B. T.; Huh, J. H.; Kron, S. J.; Mrksich, M. Nat. Biotech. 2002, 20, 270. Mitchell, P. Nat. Biotech. 2002, 20, 225. Kwon, Y.; Han, Z.; Karatan, E.; Mrksich, M.; Kay, B. K. Anal. Chem. 2004, 76, 5713. (4) Biopolymers at Interfaces; Malmsten, M., Ed.; Surfactant Science Series Marcel Dekker: New York, 1998; Vol. 75.

MCP.2,11-17 The IgG molecule has a molecular weight of 150 000 g/mole and is composed of four polypeptide chains consisting of two heavy (H) chains with 446 amino acid residues and two light (L) chains with 214 amino acid residues (Figure 1). These chains are cross linked by disulfide bonds into a flexible Y-shaped structure with dimensions of approximately 14.5 nm (height) × 8.5 nm (width) × 4 nm (thickness).18 The two upper arms of the “Y” are Fab fragments each possessing an antigen binding site (ABS) at their tips, whereas the vertical part of the “Y” represents the constant fragment (Fc). In addition to denaturing via shape deformation, the flexibility and location of the active sites of IgG (5) Andrade, J. D.; Hlady, V. AdV. Polym. Sci. 1986, 79, 1. (6) Chang, I.-N.; Lin, J.-N.; Andrade, J. D.; Herron, J. N. J. Colloids Interface Sci. 1995, 174, 10. (7) Buijs, J.; White, D. D.; Norde, W. Colloids Surf., B 1997, 8, 239. (8) Zhou, J.; Chen, S.; Jiang, S. Langmuir 2003, 19, 3472. (9) Wang, H.; Castner, D. G.; Ratner, B. D.; Jiang, S. Langmuir 2004, 20, 1877. (10) Chen, S.; Liu, L.; Zhou, J.; Jiang, S. Langmuir 2003, 19, 2859. (11) Delamarche, E. In Nanobiotechnology; Niemeyer, C. M., Mirkin, C. A., Eds.; Wiley-VCH: New York, 2004; p 31. (12) James, C. D.; Davis, R. C.; Kam, L.; Craighead, H. G.; Isaacson, M.; Turner, J. N.; Shain, W. Langmuir 1998, 14, 741; Kane, R. S.; Takayama, S.; Ostuni, E.; Ingber, D. E.; Whitesides, G. M. Biomaterials 1999, 20, 2363. Geissler, M.; Bernard, A.; Bietsch, A.; Schmid, H.; Michel, B.; Delamarche, E. J. Am. Chem. Soc. 2000, 122, 6303. Patel, N.; Bhandari, R.; Shakesheff, K. M.; Cannizzaro, S. M.; Davies, M. C.; Langer, R.; Roberts, C. J.; Tendler, S. J.; Williams, P. J. Biomater. Sci., Polym. Ed. 2000, 11, 319. Hyun, J.; Zhu, Y.; Liebmann-Vinson, A.; Beebe, T. P.; Chilkoti, A. Langmuir 2001, 17, 6358. (13) Bernard, A.; Delamarche, E.; Schmid, H.; Michel, B.; Bosshard, H. R.; Biebuyck, H. A. Langmuir 1998, 14, 2225. (14) Tan, J. L.; Tien, J.; Chen, C. S. Langmuir 2002, 18, 519. (15) Graber, D. J.; Zieziulewicz, T. J.; Lawrence, D. A.; Shain, W.; Turner, J. N. Langmuir 2003, 19, 5431. (16) LaGraff, J. R.; Zhao, Y. P.; Graber, D. J.; Rainville, D.; Wang, G. C.; Lu, T. M.; Chu-LaGraff, Q.; Szarowski, D.; Shain, W.; Turner, J. N Mater. Res. Soc. Symp. Proc. 2003, 735, 33. (17) Inerowicz, H. D.; Howell, S.; Regnier, F. E.; Reifenberger, R. Langmuir 2002, 18, 5263. (18) Silverton, E. W.; Navia, M. A.; Davies, D. R. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 5140.

10.1021/la0522303 CCC: $33.50 © 2006 American Chemical Society Published on Web 04/07/2006

4686 Langmuir, Vol. 22, No. 10, 2006

Figure 1. Schematic diagram of the Fab and Fc regions of IgG indicating the antigen binding sites (ABS). The “Y” shaped molecule is approximately 14.5 nm in height × 8.5 nm in width × 4 nm in thickness with a molecular weight of 150 000 g/mole.

make the orientation of surface-immobilized IgG (and its fragments) critical to the overall function. Protein adsorption on surfaces from solutions is a complex process and can be mediated by a number of molecular-level interactions including (1) electrostatic interactions between the protein and substrate, including coadsorption of other ions, (2) hydrogen bonding, (3) van der Waals interactions, (4) changes in the waters of hydration of the protein and the substrate, (5) conformational entropy changes due to structural rearrangement of the protein upon adsorption and, (6) lateral interactions between adsorbed protein molecules.4 The adsorption of IgG onto various materials is the subject of numerous investigations that explore the effect of solution parameters such as ionic strength, pH, chemical pretreatment of protein, and substrate charge on IgG orientation and function.4-10 IgG and its Fab and Fc fragments have measured isoelectric points of 6.8, 8.3, and 6.0, respectively.10 Although IgG is overall electrically neutral at its isoelectric point (IEP), the Fab and Fc regions would be positively and negatively charged, respectively, giving the overall IgG molecule a weak dipole capable of different orientations depending on the sign and magnitude of the surface charge, particularly at low ionic strengths where electrostatic interactions dominate.6 At high ionic strengths typical of physiological buffer solutions and the work presented here, the IgG-substrate interaction is dominated by short-range van der Waals interactions of a few nanometers or less.9 One method of patterning small molecules19-22 and proteins2,11-17 onto material surfaces is microcontact printing (MCP), a versatile, simple, and inexpensive lithographic process that uses a micro- or nanopatterned elastomeric stamp to transfer monomolecular layers of proteins onto a surface (Figure 2). MCP can produce complex patterns of stamped proteins for a diverse number of biological and bioengineering applications.11 Generally, the quality of protein pattern transfer and the maintenance of function strongly depend on how the stamp and substrate surfaces are each chemically, physically, or biochemically modified (Figure 2). For effective pattern transfer, the protein of interest not only must have a higher affinity for the substrate than the stamp but also must not bind too strongly to the substrate such that its secondary or tertiary structure deforms, causing a loss of function.23 For maximum activity, the protein molecules must attach to the substrate with their active (or binding) sites accessible to the desired target biomolecules. Both the bath application and MCP of proteins depend strongly on the substrate’s surface properties and, in the case of MCP, the surface of the PDMS stamp. Under physiologic conditions, proteins tend (19) Kumar, A.; Whitesides, G. M. Appl. Phys. Lett. 1993, 63, 2002. (20) Xia, Y.; Whitesides, G. M. Angew. Chem., Int. Ed. 1998, 37, 550. (21) Graham, D. J.; Price, D. D.; Ratner, B. D. Langmuir 2002, 18, 1518. (22) Michel, B.; Bernard, A.; Bietsch, A.; Delamarche, E.; Geissler, M.; Junker, D.; Kind, H.; Renault, J. P.; Rothuizen, H.; Schmid, H.; Schmidt-Winkel, P.; Stutz, R.; Wolf, H. IBM J. Res. DeV. 2001, 45, 697. (23) Wilson, K.; Stuart, S. J.; Garcia, A. J.; Latour, R. A. Soc. Biomat. 28th Ann. Meeting Trans. 2002, 25, 455.

LaGraff and Chu-LaGraff

to adsorb to uncharged surfaces (e.g., PDMS) through shortrange van der Waals interactions,24 although longer-range electrostatic interactions become more important at lower ionic strengths on charged surfaces.10 Proteins typically form monomolecular layers upon adsorption because multilayer formation is self-limiting as adsorbed proteins make the surface hydrophilic, thus inhibiting further protein adherence. Many groups have developed their own working MCP protocols for protein patterning; however, there is little mechanistic understanding of how the entire process occurs, particularly at the molecular level. It has been shown that the binding intensity of goat IgG to secondary antigoat IgG that had been immobilized on plasmacleaned glass substrates using MCP is 71% of that of bathapplied secondary IgG to an identically prepared substrate.15 These results and others suggest that the adherence of IgG to many types of substrates does not significantly decrease its functionality.4,11 It was shown that for MCP the most efficient transfer of IgG occurred when the substrate was more hydrophilic than PDMS stamps whose surfaces were modified with either -CF3 or -NH2 functionalities.14 This IgG surface affinity to hydrophilic substrates is the opposite of what happens during the adsorption of IgG from solution in which hydrophobic substrates tend to promote IgG adsorption.11 To understand the mechanics of IgG adsorption on surfaces better, we report a fluorescence labeling and scanning force microscopy (SFM) study that systematically explores IgG transfer mechanisms and kinetics, IgG selectivity and function, and surface chemistry as they relate to the MCP fabrication process. 2. Experimental Section Protein Preparation. The following primary and secondary antibodies and antibody fragments were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA): whole goat IgG, whole horse IgG, goat antigen binding F(ab′)2 and constant Fc IgG fragments, and tetramethyl rhodamine isothiocyanate (TRITC) conjugated secondary IgGs. Secondary IgGs included rabbit antigoat (H + L), hereafter referred to as rabbit antigoat IgG, which reacts to both the heavy and light chains of the IgG molecule, rabbit antigoat F(ab′)2 fragment-specific IgG, and rabbit antigoat Fc fragmentspecific IgG. Alexa 568 conjugated rat IgG was purchased from Molecular Probes, Inc., Eugene, Oregon. Fluorescein isothiocyanate (FITC) conjugated poly(lysine hydrobromide) (C6H12N2OHBr)n (molecular weight, 50 200 g/mol) and FITC-conjugated protein A were purchased from Sigma Chemical Company. IgG proteins were diluted to 5 ( 0.5 , 25 ( 0.5, and 50 ( 0.5 µg/mL (33.3, 166.7, and 333.3 nM, respectively) in 10 mM phosphate-buffered saline (PBS) at pH 7.4 (Sigma), which had an ionic strength of approximately 0.165 M. Protein solutions were stored at 4 °C and were used within 12 h of preparation. Substrate Preparation. Round glass coverslips (Fisher Scientific, Pittsburgh, PA) were immersed in a concentrated nitric acid solution for 24 h, rinsed in flowing deionized water for 60 min, dried for 1 h at 100 °C, and stored in a desiccator at room temperature. Five minutes prior to MCP or bath application of IgG, these acid-cleaned glass coverslips were oxygen plasma cleaned for 60 s (model PDC001, Harrick Scientific, Ossining, NY). Stamp Preparation. Patterned PDMS (Sylgard 184, Dow Corning) stamps were prepared from custom-designed topographic masters consisting of a photolithographically defined 10-µm-thick photoresist that had been spun on 3- or 4-in.-diameter silicon wafers (Silicon Quest International). PDMS is hydrophobic with advancing and receding contact angles of 115 and 95°, respectively, which facilitates the inking of the stamp with protein solutions. Stamps were stored with their patterned sides up in a covered Petri dish. Immediately prior to MCP, stamps were rinsed and ultrasonicated (24) Israelachvilli, J. Intermolecular and Surface Forces, 2nd ed.; Academic Press: London, 2000. Norde, W. AdV. Colloid Interface Sci. 1986, 25, 267-340.

Microcontact Printed Antibodies and Fragments

Langmuir, Vol. 22, No. 10, 2006 4687

Figure 2. IgG transfer to glass by microcontact printing (a-c) and bath application (d-f). (a) A PDMS stamp already inked with IgG solution (light blue) is brought into contact with a glass substrate and (b) removed to leave behind an an IgG pattern (light blue). (c) The fluorescently labeled secondary antibody (red) is then bath applied to reveal the IgG pattern shown in b. (d) A 300-µL droplet of IgG solution (light blue) is allowed to incubate on a glass substrate. (e) After rinsing in PBS, a monolayer of IgG remains. (f) After bath application and rinsing of the appropriate fluorescently labeled secondary antibody (red), one is left with a bilayer of IgG and its secondary IgG. sequentially in 50% ethanol and deionized water and then dried at 60 °C for ∼5 min. Microcontact Printing and Physical Adsorption. Details of microcontact printing of proteins have been described elsewhere (e.g., refs 15 and 16). Briefly, the MCP procedure consisted of taking a cleaned PDMS elastomeric stamp with an array of micrometer scale lines and patterns (5-25 µm widths) and incubating the stamp in 300 µL of a dilute IgG solution (5-200 µg/mL). The proteininked stamp was then rinsed in PBS, and any excess solution was removed by using a spin coater (G3P Spincoat, Specialty Coating Systems, Indianapolis, IN) for 15 s at 6500 rpm. The stamp was placed in contact with an oxygen-plasma-cleaned glass substrate for 60 min (Figure 2a). The resulting IgG pattern (Figure 2b) was incubated for 60 min in 300 µL of a 5 µg/mL TRITC-conjugated rabbit antigoat secondary antibody solution (Figure 2c). The substrates were rinsed several times after each MCP step in PBS solution to remove weakly bound biomolecules. Two final rinses of 30 min each were carried out on a platform rotator (Stovall Life Sciences Corporation, Greensboro, NC) in fresh PBS solutions. The patterned glass substrates were stored in fresh PBS at 4 °C until fluorescence microscopy or AFM imaging analysis was carried out. Physisorption of primary antibodies were carried out on similarly prepared glass substrates using 300 µL of a solution with an IgG concentration of 5, 25, or 50 µg/mL for times ranging from 30 to 150 min (Figure 2d-f) with cleaning, rinsing, and fluorescent labeling procedures identical to those used for MCP of IgG. Typically, experimental lots of 12-24 substrates at a time were fabricated as described above. Patterns appeared to be stable even after weeks of storage for samples in 4 °C PBS solution. Epifluorescence and SFM Characterization. The surface coverage of whole goat IgG was determined by measuring the fluorescence intensity of TRITC-conjugated rabbit antigoat using epifluorescence (Nikon Eclipse E600) and an external light source (X-Cite 120, Fluorescence Illumination System, Exfo Corporation, Vanier, Quebec, Canada). Images were collected at a magnification of 200× with a SPOT digital camera and software (Spot RT, Diagnostic Instruments, Inc., Sterling Heights, MI) with 6 s exposure times. For MCP substrates, background-subtracted fluorescence intensities were determined by measuring the average pixel intensity of the stamped lines using Adobe PhotoShop. At least 8 images each were collected from 3 or more substrates for each IgG solution concentration, and at least 10 patterned lines per image were measured. (Regions between MCP patterned lines were excluded from the intensity measurements.) Care was taken to avoid regions with defects (e.g., missing regions and dust particles) in the intensity analysis. For substrates with physisorbed primary IgG labeled with TRITC-conjugated secondary antibodies, at least eight images per substrate were collected from random areas on three or more substrates per IgG solution concentration in order to determine accurate pixel intensities. Occasionally other, usually longer, exposure times were used, but the intensity data was normalized to a 6 s exposure time. The fluorescence intensity was determined to be linear with exposure time. A 95% confidence limit was used to determine the error bars on all graphs. Some images showing stamping defects were obtained using an Olympus BX41 fluorescence microscope (Olympus America

Figure 3. Fluorescent micrograph of 15 µm and 20 µm wide lines of whole goat primary IgG solution (50 µg/mL) microcontact printed on glass substrate for 60 min and labeled with a TRITC-conjugated rabbit anti-goat IgG solution (5 µg/mL for 60 min). The inset shows a micrograph of physisorption of 50 µg/mL whole goat IgG solution bath applied for 60 min on identically prepared glass substrate and also labeled with the same 5 µg/mL rabbit anti-goat IgG solution. Fluorescent intensity indicates that microcontact printing transfers a higher surface concentration of IgG to a glass surface than physisorption from solution during bath applications. Constant exposure time of 6 s at a magnification of 200× was used for both micrographs. Inc.) and a Magnafire CCD camera and software. SFM imaging and analysis were carried out on a BioScope (Digital Instruments, Veeco Metrology Group) operated in tapping mode using a silicon tip with a spring constant of ∼40 N/m, radius of curvature of ∼10 nm, a typical resonance frequency in air of 300 kHz, a scan rate of 1 Hz, and 512 lines per scan. Samples were immersed in PBS solution (tip-sample forces on the order of ten’s of pico-Newtons or less). Images were stable upon repeated scanning with no obvious surface degradation or change in the molecular fine structure along the edges of stamped IgG features.

3. Results and Discussion I. MCP IgG Pattern Characterization. An array of 5-, 10-, 15-, 20-, and 25-µm-wide lines consisting of whole goat IgG (60 min, 50 µg/mL) was stamped onto glass using MCP and subsequently incubated in a TRITC-conjugated rabbit antigoat secondary IgG solution (60 min, 5 µg/mL) (Figure 3).25 For comparison, the same two IgG solutions were bath applied on an identically prepared glass substrate (Figure 3 inset). Both images were taken at a 6 s exposure time and a magnification of 200×. The relative differences in fluorescence intensity were measured in order to compare IgG surface coverage on glass substrates prepared by either bath application or MCP and as a function of concentration. (25) Time control experiments (10, 30, and 60 min) of bath applications of 5 µg/mL TRITC-labeled rabbit antigoat solutions onto bath-applied whole goat primary IgGs (50 µg/mL) showed essentially identical fluorescence intensities.

4688 Langmuir, Vol. 22, No. 10, 2006

LaGraff and Chu-LaGraff

Figure 4. 26.2 µm × 26.2 µm SFM micrograph of a 15-µm-wide MCP line consisting of whole goat IgG labeled with bath-applied TRITC-conjugated rabbit anti-goat IgG. The overall line height is 15 ( 3 nm, but the initial underlying monolayer of MCP whole goat is only 5 ( 2 nm high and can be seen around the interior of the oval pit (see arrow). Immediately prior to imaging, the sample was rinsed in distilled water and excessive droplets removed. The image was collected in PBS with a Veeco Metrology Bioscope.

Figure 4 is a 26.2 µm × 26.2 µm in-situ SFM image of a 15-µm-wide MCP stamped IgG line consisting of an IgG bilayers MCP whole goat primary IgG plus bath-applied TRITC conjugated rabbit antigoat secondary IgG. The overall line height is 15 ( 2 nm (Figure 4). The edge of the large pit (see arrow, Figure 4) shows the initial MCP monolayer of whole goat IgG with a measured height of about 5 nm. This height is consistent with X-ray crystallography measurements of IgG thickness18 and published SFM observations of other emmersed and immersed substrates consisting of a single monolayer of MCP IgG16,17 or bath-applied IgG on mica.26 After incubating the immobilized IgG monolayers in rabbit antigoat secondary IgG solution, step heights typically increased to about 15 nm, indicating that the second IgG layer thickness is around 10 nm. This suggests that whole goat primary IgG remains relatively flat on the glass substrate because of multiple areas of contact after MCP. In contrast, the bath-applied secondary rabbit antigoat IgG, which binds to the primary IgG via its antigen binding sites located at the end of the Fab regions (Figure 1), has only one or two areas of contact, which allows its Fc region to extend freely out into the solution. This is larger than the reported 9 ( 2 nm for ambient SPM imaging of an IgG bilayer reported by Inerowicz et al.17 However, the bilayer of Inerowicz et al. was fabricated by a two-step MCP process, which suggests that the second MCP stamping step also caused the second layer of IgG to lay flat on the first MCP stamped layer. The physical picture suggested by these height results is not only consistent with other SFM studies16,17,26 but also consistent with what is known about the adsorption of IgG in the dilute limit (1000 µg/mL), a large number of individual IgG molecules rapidly begin to bind with the surface.5 These IgG molecules impede each other from forming additional binding sites with the substrate, resulting in a more upright IgG conformation, a measured footprint of approximately 70 nm2, and a reported optical thickness of 32 nm.5 Optical measurements would suggest an IgG bilayer height (26) Fan, F.-R. F.; Bard, A. J. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 14222.

Figure 5. 1.14 µm × 1.14 µm SFM micrographs collected in tapping force mode (a, c, and e) and phase contrast imaging (b, d, and f) of stamped whole goat IgG labeled with TRITC-conjugated rabbit antigoat IgG as a function of whole goat solution concentration. The root-mean-square (rms) roughness of each image (a, c, and e) is 1.7, 2.2, and 3.2 nm, respectively, for whole goat solution concentrations of 5 (a and b), 25 (c and d), and 50 µg/mL (e and f). The average particle size from all three images is 21 ( 13 nm. The samples were imaged in PBS with a Veeco Metrology Bioscope.

of 28-46 nm, which is much greater than those measured by SFM.16,17,26 The experiments described here use concentrations (5-50 µg/mL) well below the dilute limit and yield SFM line height ratios consistent with IgG molecules that lie down flat when using MCP (5 nm), followed by a more upright configuration with bath application of the secondary IgG (10 nm) for a total bilayer thickness of 15 nm. To assess the details of the surfaces of MCP stamped IgG lines (e.g., Figure 3), a series of 1.14 µm × 1.14 µm in-situ tapping mode SFM images were collected from three MCP patterned samples at three different concentrations of IgG inking solution (5, 25, and 50 µg/mL) using the same imaging tip (Figure 5). Images taken at random regions show a correlation between the surface coverage of the MCP IgG protein and the root-meansquare (rms) roughness: at 5 µg/mL, the rms roughness is 1.7 nm; at 25 µg/mL, 2.2 nm; and at 50 µg/mL, 3.2 nm (Figure 5). On the basis of this random sampling, the rms roughness of these samples appears to increase with increasing whole goat IgG concentration of the ink used in MCP. Therefore, the rms roughness values are a convolution not only of the silicon tip shape but also of the whole goat IgG coverage and rabbit antigoat secondary IgG coverage. Individual features in Figure 5 are typically 20-30 nm in diameter. Particle size analysis of all three images (Figure 5a, c, and e) yielded an average particle size of 20 ( 13 nm, which is larger than the reported dimensions for

Microcontact Printed Antibodies and Fragments

Figure 6. A comparison of fluorescent intensity of whole horse and whole goat IgG baths applied to glass substrates from solution concentrations of 5, 25, or 50 µg/mL and by microcontact printing (MCP) with 50 µg/mL ink concentrations. The same rabbit anti-goat IgG solution (5 µg/mL for 60 min) was used for labeling all primary antibodies. All intensities were measured at 6-s exposure times and a magnification of 200×.

an IgG molecule (Figure 1), but may represent either tip-sample convolution effects or IgG agglomeration. II. Comparison of MCP and Bath-Applied IgG Surface Coverage. The fluorescence intensities (or surface IgG coverages) of bath-applied and MCP stamped primary IgG monolayers labeled with TRITC-conjugated rabbit antigoat secondary IgG solutions were compared to assess the number of IgG proteins adhered to the glass surface (Figure 6). Fluorescence intensities of whole horse or whole goat IgGs bath applied to glass substrates for 60 min at concentrations of 5, 25, and 50 µg/mL show an increase in fluorescent intensity (i.e., surface coverage) with increasing concentration. To determine the cross reactivity and background nonspecific binding, whole horse IgG was bath applied at identical concentrations and labeled with TRITCconjugated rabbit antigoat secondary IgG. Results show that there is almost no fluorescence intensity (Figure 6), indicating that the secondary rabbit antigoat IgG specific to whole goat IgG shows very little cross reactivity (or nonspecific binding) to primary IgGs from other animal species such as the horse. As an additional control, the nonspecific binding of the TRITCconjugated rabbit antigoat secondary IgG bath applied directly to glass substrates was also measured at concentrations of 5, 25, and 50 µg/mL, with corresponding fluorescence intensities of 3, 5, and 10 (not shown). This indicates that the nonspecific binding of rabbit antigoat secondary IgG to bath-applied whole horse IgG is even less than its nonspecific binding to the bare plasmacleaned glass substrate. A comparison of the fluorescence intensity for MCP printed whole horse primary IgG and whole goat primary IgG shows that, as expected, the rabbit antigoat secondary IgG interacts much more strongly with whole goat IgG (Figure 6). This indicates that MCP transfers a larger concentration of goat IgG to the hydrophilic glass substrate than does the bath application of goat IgG to an identically prepared glass surface. The bath application of IgG to the PDMS stamp acts to concentrate the IgG on the hydrophobic stamp surface prior to its transfer to the electrostatically charged, hydrophilic glass substrate. This electrostatic interaction between the glass substrate and IgG is stronger than the van der Waals interaction between IgG and hydrophobic PDMS, thus facilitating essentially 100% transfer. Indeed, as discussed previously, others have shown that IgG transfers more readily to a substrate that is significantly more hydrophilic than

Langmuir, Vol. 22, No. 10, 2006 4689

Figure 7. Comparison of the fluorescence intensity of physisorbed whole goat IgG as a function of bath application time (30-150 min) and concentration (5 and 50 µg/mL). Average fluorescence intensities (horizontal lines) yielded average surface coverages for each concentration. The apparent lack of a time dependence of the fluorescent intensity between 30 and 150 min indicates that equilibrium coverage is established within 30 min. All intensities were measured at 6-s exposure times at a magnification of 200×. Errors for the 5 µg/mL solutions are smaller than the symbol size.

the PDMS stamp.14 It is clear from Figure 6 that there is some cross reactivity of secondary rabbit antigoat IgG with bath-applied whole horse IgG. This suggests that if the surface concentration of whole horse IgG increases with MCP stamping then the cross reactivity and hence fluorescence intensity of the secondary IgG would also increase as shown in Figure 6. To determine fluorescence intensities of bath-applied whole goat IgG as a function of concentration (5 and 50 µg/mL) and application time, whole goat IgG was bath applied from 30 to 150 min at 25 °C (Figure 7) and then labeled with TRITCconjugated rabbit antigoat secondary IgG. The average fluorescence intensities for all times were found to be 6.4 ( 0.5 and 52.3 ( 3.1 at IgG concentrations of 5 and 50 µg/mL, respectively, which directly correlated with the equilibrium surface coverages of bath-applied IgG at the two different ink concentrations. III. Langmuir Adsorption Model. Previous studies indicate that the adsorption of proteins from a solution to a substrate has been shown to follow Langmuir-type adsorption behavior in the dilute solution limit.27 The Langmuir adsorption model for both molecules in liquids and gases assumes that only a monolayer of adsorbate is formed on the substrate, all surface sites are equivalent and can accommodate only one molecule, the adsorbate molecules do not interact with one another, and that eventually a dynamic equilibrium is established between the molecules being adsorbed and desorbed. Protein adsorption was modeled by the following Langmuir dynamic equilibrium expression

IgG(aq) + glass a IgG-glass(surface)

(1)

The equivalent two-step expression for adsorption onto a PDMS stamp during microcontact printing is represented as follows

IgG(aq) + PDMS a IgG-PDMS(surface)

(2a)

IgG-PDMS(surface) + glass f IgG-glass(surface) + PDMS (2b) (27) For example, Herron, J. N.; Wang, H. K.; Janatova, V.; Durtschi, J. D.; Caldwell, K. D.; Christensen, D. A.; Chang, I. N.; Huang, S. C. In Biopolymers at Interfaces; Malmsten, M., Ed.; Surfactant Science Series; Marcel Dekker: New York, 1998; Vol. 75, p 221.

4690 Langmuir, Vol. 22, No. 10, 2006

LaGraff and Chu-LaGraff

Figure 8. Fluorescence intensity of TRITC-conjugated rabbit antigoat secondary IgG bound to either bath-applied (() or MCP patterned (9) whole goat IgG as a function of the whole goat IgG concentration. Data from Figure 6 was used for part of the lower curve (bath applied). Fluorescence intensities were collected at 200× and measured (and some samples normalized) at a 6 s exposure time. These three dilute concentrations are probably in the linear part of the Langmuir adsorption isotherm.

where eq 2a represents the inking of the PDMS stamp and eq 2b, the complete transfer of the IgG monolayer from stamp to glass substrate. The adsorbed IgG on glass, IgG-glass(surface), is then reacted with a fluorescently conjugated secondary IgG, which allows the fractional coverage, Θ, of IgG to be related to the measured fluorescent intensity, I, by the expression

I θ) I∞

(3)

where I∞ would be the fluorescence intensity at full monolayer IgG coverage. (A value of 255 was used for I∞ because it represented the greatest luminosity value measurable with Adobe Photoshop.) The physisorption behavior of IgG can be described by the Langmuir adsorption isotherm

K[IgG] I ) I∞ 1 + K[IgG]

(4)

where K is the binding constant and [IgG] is the solution concentration of goat IgG in µg/mL. When eq 4 is rearranged, the following expression is obtained:

1 I∞ 1 +1 ) ) Θ I K[IgG]

(5)

A plot of 1/Θ (or I∞/I) versus 1/[IgG] at constant temperature will yield the binding or affinity constant, K, for a given substrateadsorbate system. The average intensities from Figure 7 were used to help construct the lower curve (“bath”) in Figure 8, which shows the average fluorescence intensity versus concentration of whole goat IgG that had been bath applied directly to the glass substrate. The upper curve (“MCP”) in Figure 8 shows the concentration dependence of the fluorescence intensity of IgG that had been bath applied to PDMS stamps and then MCP stamped onto the glass substrate. The fluorescence intensity and hence fractional monolayer coverage, Θ, of IgG increase with increasing solution concentration for both bath applications and MCP stamping. The linear intensity versus IgG concentration profile in Figure 8 suggests that the concentration range was within the linear portion of a normally parabolic Langmuir adsorption plot. Results from Figure 8 were fitted to a simple Langmuir adsorption isotherm model in order to determine binding constants, K, from eq 5. In Figure 9a, a plot of I∞/I (inverse coverage, 1/Θ) versus 1/[IgG] yielded a binding constant for IgG physisorption on glass during bath applications of Kbath ) 5.2 × 10-3 ( 4.5 × 10-4 mL/µg (or 7.8(7) × 105 M-1) (eq 5, Table 1). A Langmuir plot for MCP stamping of IgG (Figure 9b)

Figure 9. Fractional coverage, Θ, of IgG on glass is proportional to the measured fluorescence intensity of TRITC-conjugated rabbit antigoat secondary IgG bound whole goat IgG (Figure 8). Langmuir adsorption isotherms for (a) physisorption onto the glass substrate during bath applications with an IgG binding constant, Kbath ) 5.2 × 10-3 ( 4.5 × 10-4 mL/µg (7.8(7) × 105 M-1) and (b) physisorption onto PDMS stamps during inking, KMCP ) 1.1 × 10-1 ( 1.6 × 10-2 mL/µg (1.7(2) × 107 M-1). Best-fit curves had correlation coefficients of 0.99 and 0.96, respectively, for physisorption onto glass and the PDMS stamp. Table 1. Langmuir Binding Constants (Equation 5) or Affinity Constants IgG binding to...

K (M-1)

PDMS stamp during inking glass during bath application silanized glass latex avid AL gel (low ionic strength) avid AL gel (high ionic strength) antigen (avidin coated chip) antigen (carboxyl latex) protein G (avidin dextran hydrogel)

1.7(2) × 7.8(7) × 105 3(1) × 106 4.3 × 104 3.13 × 106 32.6 × 106 (0.9-1.7) × 107 (1-3) × 108 (3-8) × 108

ref 107

this work this work 6 28 29 29 30 31 32

yielded a binding constant of KMCP ) 1.15 × 10-1 ( 1.6 × 10-2 mL/µg (or 1.7(2) × 107 M-1). At each IgG concentration, a greater surface concentration, Θ, of whole goat IgG was transferred to the glass substrate by stamping with MCP than by physisorption from solution during bath applications. For comparison, several other K values are included in Table 1. MCP brings an IgG-inked PDMS stamp directly into contact with the glass substrate, where the IgG is more strongly attracted to the plasma-cleaned, hydrophilic glass substrate than the hydrophobic PDMS stamp. Therefore, one might find it somewhat surprising that the MCP process itself obeys the Langmuir adsorption model (Figure 9b). Fluorescence micrographs of PDMS stamps after MCP16,17 usually show the complete absence of IgG on raised stamped features after coming into contact with the glass substrate (Figure 2a and b). This is indicative of near 100% transfer of protein ink to the glass substrate during MCP (eq 2b). Thus, KMCP is actually a measure of the Langmuir adsorption isotherm binding constant of whole goat IgG onto the PDMS stamp during the inking of the stamp (eq 2b) and not of the actual MCP process on glass (eq 2b). Once a protein has made initial contact with the substrate, secondary conformational changes may occur, particularly in the dilute limit. Thus, although more complex two-site binding models have been proposed to account for these additional conformational changes, the authors use the simpler single-site model.4 Because nonspecific IgG adsorption to a surface such as glass is often described as irreversible especially over extended periods of time,4,5,7,11 it calls into question the exact physical meaning of a binding constant for IgG onto plasma-cleaned glass or as-prepared PDMS stamps. However, the Langmuir model offers the simplest explanation for the observed behavior of IgG

Microcontact Printed Antibodies and Fragments

Figure 10. Fluorescence intensity of various TRITC-conjugated secondary IgG’s (rabbit antigoat (H + L), rabbit antigoat F(ab′)2 fragment specific, or rabbit antigoat Fc fragment specific) that had been bath applied to various primary goat antibodies (whole goat, F(ab′)2 fragments, and Fc fragments) microcontact printed on glass.

adsorption and allows comparison with other protein adsorption studies that use the Langmuir model (e.g., Table 1).6,28-32 IV. Microcontact Printing of IgG Fragments. To assess the binding efficiency of secondary whole goat IgG antibody when compared to IgG F(ab′)2 or Fc fragments (Figure 1), MCP patterned lines were incubated in solutions of secondary antibodies specific to either whole goat IgG or its fragments, F(ab′)2 or Fc (Figure 10). Results indicated that, as expected, whole rabbit antigoat (H + L) secondary IgG reacted with the heavy and light chains of all three types of MCP stamped primary IgGs: whole goat IgG and F(ab′)2 and Fc fragments (Figure 10). When TRITCconjugated rabbit antigoat F(ab′)2 specific secondary IgG was used, it reacted with MCP patterned whole goat IgG and F(ab′)2 IgG fragments but not with MCP patterned Fc fragments (Figure 10). Finally, TRITC-conjugated rabbit antigoat Fc-specific IgG reacted only with MCP patterned whole goat IgG and Fc IgG fragments (Figure 10). As indicated by essentially zero fluorescence intensity, rabbit antigoat F(ab′)2-specific IgG did not react with MCP stamped Fc fragments, and rabbit antigoat Fcspecific IgG did not react with MCP stamped F(ab′)2 fragments (second and third data sets, respectively, in Figure 10). These results indicate that no significant loss of protein function of whole goat IgG or its fragments occurred during MCP patterning. Fluorescence microscopy (not shown) of (i) TRITC-labeled F(ab′)2 specific rabbit antigoat IgG on MCP patterned primary Fc IgG fragments and (ii) TRITC-labeled Fc specific rabbit antigoat IgG on MCP patterned primary F(ab′)2 IgG fragments both showed a higher background for the secondary IgG on the areas between the MCP patterned fragment lines because of the nonspecific binding of secondary IgG to glass. These results indicate that the MCP procedure used here was able to pattern functional antibodies and antibody fragments to glass substrates. Additional experiments are needed if one is to use fragmentspecific secondary IgGs to determine the orientation of surfaceimmobilized IgGs. V. Common Defects in MCP Patterns. Several common types of defects were observed on MCP patterned protein features including different types of pitting (e.g., Figure 4), pull outs, debris, droplets, and reverse transfer of an MCP patterned protein back to a PDMS stamp during MCP patterning of a second protein (Figure 11 and Table 2). SFM micrographs have shown debris on stamped protein features that can be attributed to either dust, (28) Taniguchi, T.; Duracher, D.; Delair, T.; Elaissari, A.; Pichot, C. Colloids Surf., B 2003, 29, 53. (29) Shi, J. Y.; Goffe, R. A. J. Chromatogr., A 1994, 686, 61. (30) Bernhard, A.; Bosshard, H. R. Eur. J. Biochem. 1995, 230, 416. (31) Ortega-Vinuesa, J. L.; Hidalgo-Alvarez, R.; de las Nieves, F. J.; Davey, C. L.; Newman, D. J.; Price, C. P. J. Colloid Interface Sci. 1998, 204, 300. (32) Polzius, R.; Diebel, E.; Bier, F.; Bilitewski, U. Anal. Biochem. 1997, 248, 269.

Langmuir, Vol. 22, No. 10, 2006 4691

salt crystals, protein agglomerates or impurities, or PDMS debris.16 For instance, excessive pitting in protein monolayers can occur because of either (i) trapped air bubbles on the substrate or PDMS stamps during bath applications or (ii) contaminants on the PDMS stamp or glass substrate (Figure 11a). These pits were formed after the bath application of 50 µg/mL whole goat IgG for 60 min followed by labeling with TRITC-conjugated rabbit antigoat IgG. Another defect observed is the pull outs (or pitting) that occurred during MCP patterning of FITC-conjugated protein A (Figure 11b). In this case, the PDMS stamp came into contact with the substrate after (or possibly before) stamping because each missing protein region on the 25-µm-wide lines is deposited immediately below each line. In contrast, excess FITCconjugated polylysine was observed to accumulate between closely spaced features such as the three differently sized window structures (Figure 11c, top of image), which are composed of 10-µm-wide lines with square openings of 10, 15, or 20 µm or between line arrays such as the 5- and 10-µm-wide lines (Figure 11c, bottom of image). These droplets were attributed to insufficient drying of the PDMS stamp after inking and prior to stamping. Finally, droplet defects were formed on 25-µm-wide MCP lines of FITC-conjugated protein A (Figure 11d). This droplet morphology is attributed to inhomogeneous drying of either the glass substrate or the PDMS stamp prior to stamping. Once the stamp is brought into contact with the glass substrate, this leaves water-rich regions of protein ink that continue to lose moisture during the 60 min stamping time, resulting in the segregation and concentration of protein. Factors that could account for stamping defects observed in Figure 11c and d are related to the degree of drying of the protein ink. Excessive polylysine remains between closely spaced features because of incomplete drying of the inked PDMS stamp (Figure 11c). Closed features such as the window structures or other closely spaced features such as the line arrays may trap excess protein ink during the drying process, leading to poor resolution of patterned features (Figure 11c). Drying of PDMS stamps for effective transfer of functional proteins is a typical optimization problem: dry the stamp too little and features will not clearly resolve; dry excessively and not only may the protein ink stay adhered to the stamp but protein that is transferred to the substrate may become denatured. Finally, there are several types of stamping defects encountered when multiple stamping of two or more proteins is carried out, such as when creating crossover arrays shown in Figure 12a.11,16,17 The fabrication of these arrays is described in more detail elsewhere,16 but they are essentially made by a two-step MCP process of two different fluorescently labeled proteins (e.g., Figure 1a and b). Figure 12b shows a fluorescence micrograph of the stamp used to create the second set of IgG lines after MCP of protein onto the first set of MCP lines. The long, dark stripes represent areas of missing IgG on PDMS stamps, which were completely transferred to the glass substrate (region 1, Figure 12b and c). A closer inspection of the PDMS stamp in Figure 12c reveals two defects commonly encountered during multiple MCP stamping on a single substrate: (i) incomplete transfer of the second protein (region 2) and (ii) reverse transfer of the first MCP protein (polylysine) to the second PDMS stamp (regions 3). Incomplete transfer of the second IgG protein occurs both on the underlying polylysine lines and on the substrate adjacent to the polylysine lines. This partial transfer of the second protein (IgG) is also supported by AFM inspection of crossover junctions as described elsewhere.16 It is important to note that aside from occasional pitting and debris features the defects described above were not frequently

4692 Langmuir, Vol. 22, No. 10, 2006

LaGraff and Chu-LaGraff

Figure 11. In addition to the pit feature observed in Figure 4, there are several other common types of MCP stamping defects found in protein monolayers, including (a) pitting, (b) pullouts, (c) excess protein between stamped features, and (d) droplet defects. In (c), FITCconjugated polylysine ink was observed to accumulate between closely spaced features such as the three different size window structures (top of image), which are composed of 10-µm-wide lines with square openings of 10, 15, or 20 µm or between line arrays such as the 5and 10-µm-wide lines (bottom of image c). Droplet defects shown in (d) can form on these 25-µm-wide MCP lines as they begin to dry during contact of the glass substrate with the inked PDMS stamp. The shadow regions are from cross-stamping of 10-µm-wide IgG lines. Images b and d are of MCP patterned FITC-conjugated protein A. The fiducial marker in a represents 100 µm. Table 2. Types of Common MCP Stamping Defects and Possible Causes and Solutions defect type

characteristics

possible causes

methods to avoid defect

debris (Figure 4 and ref 16)

debris sticking out of MCP patterned features

-dust -PBS salt crystals -protein agglomerates, impurities from protein stock solutions -PDMS debris formed during stamp fabrication

-biocleanroom -humidity control, solution chemistry -dilute protein solutions and filter sterilization -biocleanroom and stamp rinsing

Pitting (Figures 4 and 11b)

missing regions of protein

-air bubbles on stamp or on substrate -incomplete protein transfer from stamp to substrate

-surface cleanliness, degas solutions -surface cleanliness, treatment, and flatness; humidity control

merged features (Figure 11c)

excessive protein between features

-improper drying of stamp -stamp collapse

-spin coater for drying stamps -proper stamp design

droplets (Figure 11d)

circular inhomogeneities in stamped features

partial drying of features while stamp is in contact with substrate

-humidity control

reverse transfer of protein (Figure 12c)

missing regions of stamped protein after mcp stamping of a second protein

reverse transfer of already MCP stamped protein to second PDMS stamp

-solution modifications (e.g., ionic strength or additions of BSA, detergents, etc.)

encountered in the results presented in Figures 5-10. We now use a programmable spin coater to dry inked stamps reproducibly prior to MCP patterning, which has eliminated many of the defects of the type shown in Figure 11. Table 2 lists several other methods for avoiding common MCP stamping defects. Finally, it should be noted that when pits or debris features were encountered during fluorescence imaging they were excluded from the fluorescence intensity analysis.

4. Conclusions A comparison of MCP and bath-applied IgG indicated that IgG retains its function when attached to glass by either method;

however, MCP patterning results in the transfer of a greater surface concentration of IgG proteins. Measurements of the surface coverage of MCP patterned IgG versus their bath-applied counterparts onto oxygen-plasma-cleaned glass substrates yielded Langmuir binding constants of KMCP ) 1.7(2) × 107 M-1 and Kbath ) 7.8(7) × 105 M-1, respectively. The first reported screening of MCP antibody fragments described here also supported the transfer of functional IgG to the substrate. SFM characterization of MCP IgG with and without attached secondary IgG yielded line heights of approximately 5 and 15 nm, respectively. Finally, several types of MCP stamping defects were described including pitting, pull outs, droplets, and reverse

Microcontact Printed Antibodies and Fragments

Langmuir, Vol. 22, No. 10, 2006 4693

Figure 12. (a) Composite fluorescence microscopy image of protein crossover structures on glass prepared by a two-step MCP process. First, 25-µm-wide FITC conjugated polylysine (green) lines were MCP stamped onto a glass substrate (e.g., Figure 1a and b), and then 10 µm Alexa 568 conjugated whole rat IgG lines (red) were MCP stamped over the polylysine (e.g., repeat of steps a and b in Figure 1). (b) Fluorescence micrograph of the PDMS stamp after MCP patterning of rat IgG. Dark areas (region 1) indicate where most of the IgG was transferred to the substrate during stamping. The green lines in b represent the orientation and thickness of the previously stamped polylysine lines. (c) A magnified region of part b indicates regions on or near the previously stamped polylysine where IgG was partially transferred from stamp to substrate (region 2). Regions 3 indicate the reverse transfer of polylysine (green) back to the second IgG stamp.

transfer. The long-term goal of this work remains to develop a fundamental molecular-level understanding of the MCP patterning process in order to develop improved processing-structurefunction relationships for the rational design and preparation of practical protein-based devices. Acknowledgment. J.R.L. thanks Jim Turner, Gwo-Ching Wang, Song Xu, and Yi-Ping Zhao for helpful discussions and the Veeco Metrology group for SFM images shown in this article. Brent Matteson, Margo E. Rockwell, Anna Arnold, and Brian

Clark are acknowledged for assistance with sample preparation and characterization. J.R.L. acknowledges support from the Research Corporation (CC5751) and the American Chemical Society Petroleum Research Fund (38872-GB5S). Work was performed in part at the Cornell Nanofabrication Facility (a member of the National Nanofabrication Users Network), which is supported by the National Science Foundation under grant ECS-9731293. LA0522303