Langmuir 2004, 20, 11123-11126
11123
Protein Surface Patterning Using Nanoscale PEG Hydrogels Ye Hong, Peter Krsko, and Matthew Libera* Department of Chemical, Biomedical, and Materials Engineering, Stevens Institute of Technology, Hoboken, New Jersey 07030 Received June 1, 2004. In Final Form: August 13, 2004 We have used focused electron-beam cross-linking to create nanosized hydrogels and thus present a new method with which to bring the attractive biocompatibility associated with macroscopic hydrogels into the submicron length-scale regime. Using amine-terminated poly(ethylene glycol) thin films on silicon substrates, we generate nanohydrogels with lateral dimensions of order 200 nm which can swell by a factor of at least five, depending on the radiative dose. With the focused electron beam, high-density arrays of such nanohydrogels can be flexibly patterned onto silicon surfaces. Significantly, the amine groups remain functional after e-beam exposure, and we show that they can be used to covalently bind proteins and other molecules. We use bovine serum albumin to amplify the number of amine groups, and we further demonstrate that different proteins can be covalently bound to different hydrogel pads on the same substrate to create multifunctional surfaces useful in emerging bio/proteomic and sensor technologies.
Introduction The binding of proteins and other biomolecules to surfaces is essential to a variety of applications, including high-throughput proteomic arrays, directed cell adhesion, and biosensors.1-6 These applications present many challenging materials-design criteria. Prominent among them is the need to increase the areal density of protein binding sites on a surface for enhanced throughput and efficiency of multianalyte proteomic assays, as well as the need to control cell/surface interactions at subcellular length scales. A number of patterning technologiesssoft lithography,4,7,8 dip-pen nanolithography and other scanned-probe patterning methods,9-13 focused-ion-beam nanopatterning,14 and direct-write electron-beam lithography,15,16 among othersshave successfully demonstrated protein localization at ∼10-nm to 10-µm length scales. In * Author to whom correspondence should be addressed. E-mail:
[email protected]. (1) Constans, A. The Scientist 2002, 16 (9), 28. (2) Chen, C. S.; Mrksich, M.; Huang, S.; Whitesides, G. M.; Ingber, D. E. Science 1997, 276, 1425-1428. (3) MacBeath, G.; Schrieber, S. L. Science 2000, 289 (8 Sept.), 17601763. (4) Tan, J. L.; Tien, J.; Chen, C. S. Langmuir 2002, 18, 519-523. (5) Templin, M. F.; Stoll, D.; Schrenk, M.; Traub, P. C.; Vohringer, C. F.; Joos, T. O. Trends Biotechnol. 2002, 20 (4), 160-166. (6) Yadavalli, V. K.; Koh, W.-G.; Lazur, G. J.; Pishko, M. V. Sensors Actuators 2004, B97, 290-297. (7) Kane, R. S.; Takayama, S.; Ostuni, E.; Inber, D. E.; Whitesides, G. M. Biomaterials 1999, 20, 2363-2376. (8) Michel, B.; Bernard, A.; Bietsch, A.; Delamarche, E.; Geisssler, M.; Juncker, D.; Kind, H.; Renault, J.-P.; Rothiuzen, H.; Schmid, H.; Schmidt-Winkel, P.; Stutz, R.; Wolf, H. IBM J. Res. Dev. 2001, 45 (5), 697-719. (9) Kenseth, J.; Harnisch, J. A.; Jones, V. W.; Porter, M. D. Langmuir 2001, 17, 4105-4112. (10) Lee, K.-B.; Park, S.-J.; Mirkin, C. A.; Smith, J. C.; Mrksich, M. Science 2002, 295 (1 March), 1702-1705. (11) Lee, K.-B.; Lim, J.-H.; Mirkin, C. A. J. Am. Chem. Soc. 2003, 125, 5588-5589. (12) Liu, G.-Y.; Xu, S.; Qian, Y. Acc. Chem. Res. 2000, 33, 457-466. (13) Wadu-Mesthrige, K.; Xu, S.; Amro, N. A.; Liu, G.-Y. Langmuir 1999, 15, 8580-8583. (14) Bergman, A. A.; Buijs, J.; Herbig, J.; Mathes, D. T.; Demarest, J. J.; Wilson, C. D.; Reimann, C. T.; Baragiola, R. A.; Hull, R.; Oscarsson, S. O. Langmuir 1998, 14, 6785-6788. (15) Mendes, P. M.; Jacke, S.; Critchley, K.; Plaza, J.; Chen, Y.; Nikitin, K.; Palmer, R. E.; Preece, J. A.; Evans, S. D.; Fitzmaurice, D. Langmuir 2004, 20, 3766-3768.
addition to localizing proteins, however, clinically relevant applications require that protein conformation be maintained in order to effectively display natural function. In contrast to physisorption or protein tethering directly to a hard surface, the soft and hydrated environment characteristic of a swollen hydrogel can present nearphysiological conditions that minimize denaturation and are thus attractive for displaying functional proteins. In fact, there is little thermodynamic driving force for protein adsorption on many hydrogel surfaces, and the proteinrepelling properties of, for example, PEG hydrogels and PEGylated surfaces are well established.17-21 Proteins can be covalently bound to hydrogel surfaces where they can signal for cell adhesion22-25 or for analyte binding,26,27 but a challenge has remained in trying to pattern proteinfunctional hydrogels at submicron length scales. Here, we report a method to create surface-patterned functional hydrogels with nanoscale dimensions, and we thus simultaneously increase the areal density of bound proteins while maintaining a highly hydrated binding surface. We use a focused electron beam to locally crosslink dry poly(ethylene glycol) (PEG) thin films cast on silicon substrates. A focused electron beam provides a (16) Stamou, D.; Musil, C.; Ulrich, W.-P.; Leufgen, K.; Padeste, C.; David, C.; Gobrecht, J.; Duschl, C.; Vogel, H. Langmuir 2004, 20, 34953497. (17) Revzin, A.; Tompkins, R. G.; Toner, M. Langmuir 2003, 19, 98559862. (18) Sofia, S. J.; Premnath, V.; Merrill, E. W. Macromolecules 1998, 31, 5059-5070. (19) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, 10714-10721. (20) Gombotz, W. R.; Wang, G.; Horbett, T. A.; Hoffman, A. S. J. Biomed. Mater. Res. 1991, 25, 1547-1562. (21) Alcantar, N. A.; Aydil, E. S.; Israelachvili, J. N. J. Biomed. Mater. Res. 2000, 51, 343-351. (22) Kao, W. J.; Hubbell, J. A. Biotechnol. Bioeng. 1997, 59 (1), 2-9. (23) Kao, W. J. Biomaterials 1999, 20, 2213-2221. (24) Gobin, A. S.; West, J. L. Biotechnol. Prog. 2003, 19 (6), 17811785. (25) Drumheller, P. D.; Hubbell, J. A. Anal. Biochem. 1994, 222, 380-388. (26) Guschin, D.; Yershov, G.; Azaslavsky, A.; Gemmell, A.; Shick, V.; Proudnikov, D.; Arenkov, P.; Mirzabekov, A. Anal. Biochem. 1997, 250, 203-211. (27) Arenkov, P.; Kukhtin, A.; Gemmell, A.; Voloshchuk, S.; Chupeeva, V.; Mirzabekov, A. Anal. Biochem. 2000, 278, 123-131.
10.1021/la048651m CCC: $27.50 © 2004 American Chemical Society Published on Web 11/02/2004
11124
Langmuir, Vol. 20, No. 25, 2004
means to deliver energy to a specimen with very high spatial resolution, and the electrostatic beam-positioning capabilities provided by modern electron-optical systems enables that energy to be delivered to very specific locations and in user-defined patterns. We have previously reported on the fabrication of micron-scale hydroxy-terminated PEG 6800 gels patterned on silicon, their swelling properties as a function of dose-controlled cross-link density, and their affinity to differentially repel or adsorb fibronectin, a cell-adhesion-promoting protein found in the extracellular matrix (ECM).28 Now, we extend this work to address a series of questions relevant to nanoscale protein surface patterning. Using commercially available monoamine-terminated PEG 5000, we demonstrate that the e-beam approach can be used to make ∼200-nm diameter nanohydrogels. We confirm that the amine groups survive the electron-beam irradiation and then amplify the number of functional amines to increase the binding-site density. Finally, we immobilize different proteins on different hydrogel pads on the same silicon substrate in a manner consistent with multianalyte protein chip design. Experimental Procedure
Hong et al.
Figure 1. AFM images of a 5 µm × 5 µm array of amineterminated PEG 5000 nanohydrogels with an inter-gel spacing of 715 nm. (A) dry; (B) hydrated; (C) height profiles of a row of nanogels which swell by a factor exceeding five times. pad was exposed to BSA at the same concentration but with no EDAC. Both samples were kept at room temperature for 2 h and then washed repeatedly with 0.1 M PBS and then 1 M NaCl buffer solution. The pads were subsequently allowed to react with FITC solution overnight at 4 °C as described above. To create gels which could be selectively functionalized with different proteins, we generated arrays of photoactivatable nanohydrogels using a water-soluble heterobifunctional crosslinkerssulfosuccinimidyl-6-[4′-azido-2′-nitrophenylamino]hexanoate (sulfo-SANPAH).29 To link its sulfo-NHS ester group to bound BSA amines, sulfo-SANPAH dissolved in water was pipetted onto wafers, each patterned with four array pads, and allowed to react for 30 min under dark conditions. The nitrophenyl azide on the other end of crosslinker was then available to bind amines on target proteins under UV exposure (320-350 nm). We spotted laminin (Ln) onto one array pad, fibronectin (Fn) onto another, and PBS onto two others to serve as controls. The entire wafer was then exposed to UV light and subsequently washed. To confirm that the different pads were functionalized with different proteins, the wafers were exposed to a solution containing mixed primary antibodies of anti-rabbit Ln and antihuman Fn (30 min, 20 °C) followed by normal donkey serum (NDS) blocking. The wafers were then exposed to a solution of mixed secondary antibodies (30 min, 20 °C). We used donkey anti-rabbit IgG-FITC (495 nm/525 nm) with minimum crossreaction to rabbit and donkey anti-human IgG-TR (Texas Red, 596 nm/620 nm) with minimum cross-reaction to human (Jackson ImmunoResearch). The specimens were washed, dried using nitrogen, and imaged to determine both the FITC and TR intensities.
We used monoamine-terminated PEG 5000 (PEG-NH2) or hydroxy-terminated PEG 6800 (PEG 6800) (Fluka) and methods described previously28 to create nano- and micro-sized surface hydrogels. Briefly, 1 wt% PEG solutions in THF were spin cast on plasma-cleaned silicon wafers pretreated by immersion in n-(triethoxysilylpropyl)-o-poly(ethylene oxide) urethane. The films were exposed to a 10-keV focused electron beam using a modified LEO 982 field-emission scanning electron microscope (FEG-SEM). Using an external computer system interfaced to the microscope, we could conveniently vary both the dwell time of the electron beam at each pixel position, as well as the interpixel spacing between adjacent irradiation points. For a given beam current, we varied the pixel dwell time to sample radiative doses ranging from 0.95 to 234 C/m2. We also varied the interpixel spacing from 1250 nm down to 50 nm. After irradiation, the samples were thoroughly rinsed in THF and then in water to remove unexposed or insufficiently cross-linked polymer. To image the patterned hydrogels, we used a Digital Instruments NanoScope IIIa atomic force microscope (AFM) operated in contact mode. The imaging force was minimized in order to limit deformation of the nanohydrogels by the AFM tip. We used a flourescein isothiocyanate (FITC) assay to show that the amine endgroups remain chemically active after irradiation. We created a series of hydrogel array pads, each with overall lateral dimensions of 5 µm × 5 µm, formed with a 250-nm interpixel spacing using a pixel dwell time of 500 µs and a beam current of 0.39 pA to give a dose of 2.34 C/m2 where the gels swell by a factor of approximately 2. At an interpixel spacing of 250 nm, the individual nanohydrogels cannot be resolved using conventional optical microscopy and the array appears like a continuous hydrogel. We explored interpixel spacings as small as 50 nm, where the nanohydrogels do, in fact, overlap to make a continuous gel structure over the 5 µm × 5 µm area. The various samples were exposed to a FITC solution (0.5 mg in 1 mL sodium carbonate buffer) and held at 4 °C overnight to form thiourea bonds. Unreacted FITC was removed by repeated washing with buffer. If active amine groups exist, the hydrogels should fluoresce green. We imaged these specimens using a Nikon Eclipse E1000 fluorescence optical microscope, and we used Image Pro Plus software to quantify the resulting digital fluorescence images. To amplify the concentration of the amine groups associated with the PEG-NH2, we covalently bound BSA to them using carbodiimide chemistry. For this experiment, a hydrogel array pad was exposed to a solution containing 1 mg/mL of BSA together with freshly prepared 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDAC). As a control, an identical array
Similar to our previous experiments on PEG 6800, we found that stable surface-patterned hydrogels could be created from PEG-NH2 using incident doses ranging from 0.95 to 234 C/m2, and we observed that the degree of crosslinking increases with electron dose. In contrast to our previous studies, however, which concentrated on gel pads with lateral dimensions of approximately 5 µm × 5 µm, here, we explored whether we could create discrete hydrogels with nanoscale dimensions. We found that this is indeed the case. Figure 1 shows AFM images of a 5 µm × 5 µm array of 49 individual hydrogels in both the dry (1A) and hydrated (1B) states. These were created using an incident beam approximately 10 nm in diameter with
(28) Krsko, P.; Mansfield, M.; Sukhishvili, S.; Libera, M.; Clancy, R. Langmuir 2003, 19, 5618-5625.
(29) Hermanson, G. T. Bioconjugate Techniques; Academic Press: San Diego, 1996.
Results and Discussion
Protein Surface Patterning Using Nanoscale PEG Hydrogels
Figure 2. Fluorescent intensities from FITC-labeled gel pads (250-nm interpixel spacing) showing that amines remain functional after e-beam irradiation and that they can be amplified with BSA. (A) shows no autofluorescence from PEGNH2 5000 gels; (B) shows no autofluorescence from PEG 6800 gels; (C) shows that FITC binds to PEG-NH2 5000 gels; (D) shows no binding of FITC to e-beam cross-linked PEG 6800 gels indicating that the binding in (C) is due to the amine groups; (E) FITC binding to BSA amplified PEG-NH2 5000 gels; (F) shows that BSA physisorption is small.
single point irradiations separated from each other by 715 nm. The beam current for the experiment was 0.078 pA, and the dwell time at each pixel position was 125 µs. Height profiles (Figure 1C) indicate that these gels swell vertically from a dry height of approximately 24 nm to a hydrated height of approximately 125 nm, leading to a swell ratio exceeding 5. When we increase the pixel dwell time to 209 µs (result not shown), the swell ratio decreases to 2.8, indicating an increased cross-link density consistent with our measurements on PEG 6800.28 We attribute the fact that the lateral nanogel dimension, approximately 170 nm, is substantially larger than the incident beam size in part to the proximity effect,30-32 where electrons backscattered by the substrate traverse the polymer film a second time at points up to hundreds of nanometers from the incident electron beam, as well as to electron scattering within the polymer itself.33 The lateral size of the nanohydrogels decreases significantly when we use a thin-film substrate to reduce the number and spatial extent of proximity-effect electrons. One can see from Figure 1C that the swelling occurs principally in one dimension, similar to our previous studies of microhydrogel pads. Having demonstrated that stable nanohydrogels can be formed by focused electron-beam cross-linking, we next investigated whether the amine endgroups remain chemically active after irradiation using an FITC assay. Figure 2 summarizes the result. Gel pad C was formed from PEGNH2 and then exposed to a FITC solution overnight to form thiourea bonds. As controls to demonstrate the absence of autofluoresence from the polymer, gel pads A and B were formed from PEG-NH2 and PEG 6800, respectively, but not exposed to a FITC. As a control to demonstrate that the amine endgroup is responsible for the FITC binding, gel pad D was formed from PEG 6800 and exposed to FITC at the same time as pad C. Unreacted (30) Murata, K.; Kyser, D. F.; Ting, C. H. J. Appl. Phys. 1981, 52 (7), 4396-4405. (31) Rai-Choudary, P. Handbook of Microlithography, Micromachining, and Microfabrication; SPIE: Bellingham, 1997; Vol. 1. (32) Thompson, L. F.; Willson, C. G.; Bowden, M. J. Introduction to Microlithography, 2nd ed.; American Chemical Society: Washington, DC, 1994. (33) Siangchaew, K.; Libera, M. Philos. Mag. A 2000, 80 (4), 10011016.
Langmuir, Vol. 20, No. 25, 2004 11125
Figure 3. The fluorescent intensity emitted by FITCconjugated PEG-NH2 nanohydrogel arrays collected from 5 µm × 5 µm areas for various interpixel spacings.
FITC was removed by repeated washing with buffer. If amine groups exist and are responsible for the FITC binding, gel pad C should fluoresce green while pads A, B, and D should remain dark. Figure 2 indeed demonstrates that this is the case. Using Image Pro Plus software to quantify digital fluorescence images, we found that the intensity of gel pad C is 7.5 times greater than that of each of the three control gel pads, indicating that the NH2 groups indeed remain chemically active and can be used as sites for covalent grafting. We observed similar behavior in gel pads exposed to electron doses as high as 95 C/m2. To demonstrate that functionalization is possible at the length scale of an individual nanohydrogel, we created arrays of either independent or overlapping nanohydrogels by varying the interpixel spacing while fixing the electron beam dwell time at each pixel position. Using digital fluorescent images from covalently bound FITC, we measured the fluorescent intensity over an area of 5 µm × 5 µm for each of nine different interpixel spacings between 50 and 1250 nm. We repeated the experiment five times on two different silicon substrates. Interestingly, we found that the intensity is constant for interpixel spacings of 300 nm and below (Figure 3) despite the fact that the total dose (number of electrons per unit area) increases by a factor of 35 when decreasing the interpixel spacing from 300 to 50 nm. This finding suggests that the total number of accessible amine groups is roughly constant over a wide range of electron-irradiation conditions once the nanohydrogels begin to overlap and form laterally continuous structures. In addition, the fact that the five different experiments provide raw intensity values which vary over a range less than 10% of the average suggests a level of reproducibility sufficient to make robust devices. Since the absolute intensity in Figure 3 decreases as the interpixel spacing increases beyond 300 nm, the nanohydrogels must be sufficiently separated that they do not overlap. An increasing interpixel spacing simply means that there is less fluorescing material per unit area. This fact is borne out clearly by Figure 4 which shows a fluorescence micrograph of nanohydrogel arrays for interpixel spacings of 1250, 715, 500, and 300 nm together with profiles of the fluorescent intensity across one row from each of the four arrays (Figure 4B). Strikingly, individual nanohydrogels can be clearly distinguished on the basis of the fluorescent signal from the covalently bound FITC despite the fact that each nanohydrogel is itself too small to be resolved in a conventional optical microscope. The intensity profiles show that the peak fluorescent signal emitted by a single nanohydrogel is
11126
Langmuir, Vol. 20, No. 25, 2004
Figure 4. (A) Fluorescent image of nanohydrogel arrays with different inter-gel spacings with FITC covalently bound monoamine-terminated PEG 5000 nanohydrogels; (B) Profile of fluorescent intensity from a single row of nanohydrogels at various interpixel spacings.
comparable to that of the overlapping arrays (interpixel spacings of 300 nm and below). The fact that there is relatively little peak-to-peak variation of the fluorescent intensity emitted by the individual nanohydrogels again suggests a level of reproducibility sufficient to create robust devices. Recognizing the low amine concentration intrinsic to the monoamine-terminated PEG-NH2, we amplified this concentration by covalently binding BSA. Dendrimers or similar molecules with a high surface concentration of multifunctional groups could alternately be used as amplification agents. The results are presented in Figure 2 for pads E and F. Pad E was formed under identical irradiation conditions as Pad C but then also exposed to a solution containing both BSA and EDAC. As a control, pad F was processed identically to pad E but was never exposed to EDAC. Both pads E and F were rinsed in highsalt buffer in order to remove physisorbed BSA. These pads were subsequently allowed to react with FITC, and Figure 2 shows that pad E exhibits a fluorescent intensity more than four times greater than that of the unamplified PEG-NH2 pad (pad C), indicating that BSA has covalently linked to NH2 groups on pad C via the carbodiimide reaction.29 The intensity of pad F is approximately the same as that of pad B, indicating that the extent of BSA physisorption is low. We finally addressed the issue of how different pads on the same substrate might be functionalized with different proteins. Multifunctionality is important in protein-chip technology for multianalyte assays and in cell-chip technology for co-culture applications. In current practice, photolithography or spotting is used to impart spatially resolved functionality to a chip surface. Here, we present results using microscale spotting to functionalize PEGNH2 hydrogel pads on the same substrate with Fn or with Ln. To take advantage of the nanosized dimensions possible with electron-beam patterned nanogels, however, one could apply the same principles using a nanoscale spotting technology such as dip-pen nanolithography.10,11 For this experiment, we created BSA-amplified PEGNH2 gel pads with lateral dimensions of 5 µm × 5 µm using a 100-nm interpixel spacing, and we linked a photoactivatable heterobifunctional cross-linker (sulfoSANPAH)29 to bound BSA amines. We spotted Ln onto a specific gel pad (C1), Fn onto another (D1), and PBS onto pads A1 and B1 as controls. The entire wafer was then exposed to UV light in order to bind the various proteins to the gel pads via the exposed nitrophenyl azide end of the sulfo-SANPAH. Using the immunofluorescence assay described in our experimental procedure section, we studied the differential binding of Ln and Fn to their respective gel pads by following the FITC and TR intensities. Results from the spotting experiment are plotted in Figure 5. The average background signals due
Hong et al.
Figure 5. FITC and TR tagged secondary antibodies for Fn and Ln, respectively, indicate that different gel pads can be effectively labeled with different probe proteins: (A1) control; (B1) control; (C1) pad with bound Fn; (D1) pad with bound Ln.
to nonspecific binding of Fn and Ln to the silicon surrounding the functionalized gel pads are 148 counts and 208 counts, respectively, and Figure 5 presents intensity values after subtraction of these background values. Figure 5 shows that the differential ratio of the assay and control signals, Ra/c, for Ln is 8.1 and that for Fn is 9.1, indicating that Ln and Fn are selectively attached to different gel pads on the same substrate with high differentiation. Conclusion We have extended previous experiments on e-beam cross-linked hydroxy-terminated PEG 6800 to show that dry amine-terminated PEG 5000 thin films can similarly be cross-linked to create nanosized hydrogels on silicon substrates. The amine endgroup introduces functionality absent in our previous work, and we have shown that these amine groups can be used to covalently bind proteins to highly hydrated nano and microhydrogels. Because of the hydration, we expect that covalently bound protein will present itself in a manner consistent with an in vivo environment. These electron-beam-cross-linked thin-film hydrogels can be made smaller than 200 nm in lateral size, and when functionalized with FITC, fluorescence microscopy can image individual nanohydrogels. Given the flexibility of the electron-beam approach, they can be patterned as isolated features or in high-density arrays, depending on the application. Because of their small size, the areal density of information in a proteomic or genomic array can thus potentially be increased by as much as 4 orders of magnitude. Interesting questions remain regarding the extent to which one can control the degree of grafting to a specific nanohydrogel, and this is an area we continue to study. In sum, these nanosized functional PEG hydrogels present a new means to bring the attractive biocompatibility associated with macroscale hydrogels into the submicron length-scale regime. Acknowledgment. The authors gratefully acknowledge RESBIO-The National Resource For Polymeric Biomaterials (NIH NIBIB Grant No. PH1 EB00104601A1), the Army Research Office (ARO Grant No. DAAD19-03-1-0271), and for support of electron microscopy instrumentation at Stevens, the National Science Foundation. LA048651M
