pubs.acs.org/Langmuir © 2010 American Chemical Society
Carbohydrate Microarrays by Microcontact Printing Christian Wendeln,† Andreas Heile,‡ Heinrich F. Arlinghaus,‡ and Bart Jan Ravoo*,† †
Organic Chemistry Institute and Center for Nanotechnology (CeNTech), Westf€ alische Wilhelms-Universit€ at M€ unster, Corrensstrasse 40, 48149 M€ unster, Germany, and ‡Physikalisches Institut, Westf€ alische Wilhelms-Universit€ at M€ unster, Wilhelm-Klemm-Strasse 10, 48149 M€ unster, Germany Received September 21, 2009. Revised Manuscript Received November 13, 2009
This Article describes the preparation of carbohydrate microarrays by the immobilization of carbohydrates via microcontact printing (μCP) on glass and silicon substrates. To this end, diene-modified carbohydrates (galactose, glucose, mannose, lactose, and maltose) were printed on maleimide-terminated self-assembled monolayers (SAMs). A Diels-Alder reaction occurred exclusively in the contact area between stamp and substrate and resulted in a carbohydrate pattern on the substrate. It was found that cyclopentadiene-functionalized carbohydrates could be printed within minutes at room temperature, whereas furan-functionalized carbohydrates required long printing times and high temperatures. By successive printing, microstructured arrays of up to three different carbohydrates could be produced. Immobilization and patterning of the carbohydrates on the surfaces was investigated with contact angle measurements, X-ray photoelectron spectroscopy (XPS), time-of-flight secondary ion mass spectrometry (TOF-SIMS), and fluorescence microscopy. Furthermore, the lectins concanavalin A (ConA) and peanut agglutinin (PNA) bind to the microarrays, and the printed carbohydrates retain their characteristic selectivity toward these proteins.
Introduction It is widely recognized that, among the various biological polymers, carbohydrates have an extraordinary position. In contrast to nucleic acids and proteins, carbohydrates not only have the possibility to form linear oligomers and polymers but they can also form highly branched structures. Furthermore, many different glycosidic connections between the monosaccharide units are found in nature, which give rise to structures of high complexity and diversity.1,2 In this sense, carbohydrates contain much more structural information than nucleic acids or proteins. In view of the complexity of biological carbohydrate structures, one might comprehend that our understanding of the cellular processes that carbohydrates mediate is still lagging far behind our knowledge about proteins and nucleic acids.3 In addition, glycobiology has suffered from a lack of analytical and synthetic tools which have been established as standard methods in protein and nucleic acid research, such as automated sequencing, automated synthesis, and high-throughput microarray screening.4 Nevertheless, these techniques have also been adapted to carbohydrate research during the last years and are achieving increasing biomedical relevance.5-7 In particular, carbohydrate microarrays are developing rapidly since their first description in 2002 by several groups.8-11 Carbohydrate microarrays usually consist of a flat surface on which different carbohydrates are immobilized in a well-defined and ordered microarray format. Among others, carbohydrate microarrays can be used for the *To whom correspondence should be addressed. E-mail: b.j.ravoo@ uni-muenster.de. (1) Dwek, R. A. Chem. Rev. 1996, 96, 683. (2) Hurtley, S.; Service, R.; Szuromi, P. Science 2001, 291, 2337. (3) Feizi, T.; Chai, W. Nat. Rev. Mol. Cell Biol. 2004, 5, 582. (4) Seeberger, P. H.; Werz, D. W. Nature 2007, 446, 1046. (5) Venkataraman, G.; Shiver, Z.; Sasisekharan, R. Science 1999, 256, 537. (6) Seeberger, P. H. Chem. Soc. Rev. 2008, 37, 19. (7) Park, S.; Lee, M.; Shin, I. Chem. Commun. 2008, 4389. (8) Houseman, B. T.; Mrksich, M. Chem. Biol. 2002, 9, 443. (9) Park, S.; Shin, I. Angew. Chem., Int. Ed. 2002, 41, 3180. (10) Fukui, S.; Feizi, T.; Galustian, C.; Lawson, A. M.; Chai, W. Nat. Biotechnol. 2002, 20, 1011. (11) Houseman, B. T.; Gawalt, E. S.; Mrksich, M. Langmuir 2003, 19, 1522.
Langmuir 2010, 26(7), 4933–4940
analysis of carbohydrate-protein interactions, to determine substrate specificity of carbohydrate processing enzymes, to detect interactions with antibodies, or to characterize carbohydratemediated cell recognition events.12 Carbohydrate microarrays have, for example, been used to study the binding affinity of HIV-1 glycoproteins to different glycans,13 to investigate glycan binding properties of human T-cells,14 and to screen fucosyltransferase inhibitors.15 Also, interactions of heparin binding proteins to microarrays of synthetic heparin oligosaccharides have been investigated.16 The immobilization of carbohydrates is critical to the preparation of a carbohydrate microarray. The various approaches that have been demonstrated for the surface immobilization of carbohydrates can be divided in noncovalent and covalent attachment strategies. In the case of noncovalent attachment, the carbohydrates adhere to the surface by adsorption. Covalent immobilization relies on a chemical reaction that takes place between a chemically functionalized surface and a carbohydrate derivative and therefore allows the fabrication of high stability arrays.17 Diels-Alder reactions are very suitable for covalent immobilization of biomolecules, since they are compatible and orthogonal to a large variety of chemical groups and they are fast and quantitative, provided a suitable combination of diene and dienophile is selected.18-21 Also, Diels-Alder reactions can be (12) Shin, I.; Park, S.; Lee, M. Chem.;Eur. J. 2005, 11, 2894. (13) Adams, E. W.; Ratner, E. W.; Boeksch, H. R.; McMahon, B. R.; O’Keefe, B. R.; Seeberger, P. H. Chem. Biol. 2004, 11, 875. (14) Nimrichter, L.; Gargir, A.; Gortler, M.; Alstock, A.; Shtevi, A.; Weisshaus, O.; Fire, E.; Doton, N.; Schnaar, R. L. Glycobiology 2004, 14, 197. (15) Bryon, M. C.; Lee, L. V.; Wong, C.-H. Bioorg. Med. Chem. Lett. 2004, 14, 3185. (16) De Paz, J. L.; Noti, C.; Seeberger, P. H. J. Am. Chem. Soc. 2006, 128, 2766. (17) Monzo, A.; Guttman, A. QSAR Comb. Sci. 2006, 25, 1033. (18) Husar, G. M.; Anziano, D. J.; Leuck, M.; Sebesta, D. P. Nucleosides, Nucleotides Nucleic Acids 2001, 20, 559. (19) Houseman, B. T.; Huh, J. H.; Kron, S. J.; Mrksich, M. Nat. Biotechnol. 2002, 20, 270. (20) Dillmore, W. S.; Yousaf, M. N.; Mrksich, M. Langmuir 2004, 20, 7223. (21) De Araujo, A. D.; Palomo, J. M.; Cramer, J.; K€ohn, M.; Schr€oder, H.; Wacker, R.; Niemeyer, C.; Alexandrov, K.; Waldmann, H. Angew. Chem., Int. Ed. 2006, 45, 296.
Published on Web 01/21/2010
DOI: 10.1021/la903569v
4933
Article
conducted in aqueous environment under mild conditions. Seminal work concerning the immobilization of carbohydrates using Diels-Alder chemistry was done by Houseman and Mrksich, who prepared carbohydrate chips by the immobilization of carbohydrate-cyclopentadiene conjugates on self-assembled monolayers that presented benzoquinone and penta(ethylene glycol) groups.8 The benzoquinone was generated on the surface by electrochemical oxidation of hydroquinone. The penta(ethylene glycol) chains prevented the nonspecific absorption of protein. Microarrays fabricated by this method were shown to be useful for profiling the binding specifities of lectins and to determine the inhibitory concentrations of soluble carbohydrates. This could be shown by using fluorescence microscopy and surface plasmon resonance spectroscopy (SPR). Another approach for DielsAlder-mediated immobilization of carbohydrates was presented by Sun et al., who reacted cyclopentadiene-modified lactose with a maleimide-functionalized glass surface.22 Cyclopentadiene is very reactive toward maleimide.23,24 Carbohydrate microarrays are usually fabricated by microspotting carbohydrate probes on activated substrates. This can be done with a robotic microarray printer, similar to the state-ofthe-art preparation of DNA microarrays,25,26 leading to high density chips with a huge amount of spots, which can be conveniently analyzed by an array scanner using fluorescence microscopy. Disadvantages of this method are that the resolution of printing is limited, the probe density is not easily reproducible, and that the spots tend to be very heterogeneous. In this study, we exploit the Diels-Alder reaction of cyclopentadiene-modified and furan-modified carbohydrates on maleimide self-assembled monolayers (SAMs) to immobilize monoand disaccharides on silicon and glass. In contrast to the approaches outlined above, the Diels-Alder reaction was locally induced on the surface by microcontact printing (μCP) of carbohydrate inks on an intermediate SAM, that is, by “microcontact chemistry”.27 μCP28-30 is an emerging replication method for biological microarrays, including protein31,32 and DNA microarrays.33-36 Recently, we presented the first carbohydrate microarray made by “click chemistry” of azides and alkynes induced by μCP.37 Other groups have also started to use μCP for the preparation of carbohydrate microarrays.38 A significant advantage of Diels-Alder reactions compared to “click chemistry” is that, in Diels-Alder chemistry, a Cu(I) catalyst is not required. It is hard to remove this catalyst from the carbohydrate surface, and it can interfere with protein-carbohydrate interactions in the microarray. Advantages of μCP compared to microarray fabrication by (22) Sun, X.; Yang, L.; Chaikof, E. L. Tetrahedron Lett. 2008, 49, 2510. (23) Sauer, J.; Lang, D.; Mielert, A. Angew. Chem., Int. Ed. Engl. 1962, 5, 268. (24) Sauer, J. Angew. Chem., Int. Ed. Engl. 1966, 5, 211. (25) Heise, C.; Bier, F. F. Top. Curr. Chem. 2006, 261, 1. (26) Pirrung, M. C. Angew. Chem., Int. Ed. 2002, 41, 1276. (27) Ravoo, B. J. J. Mater. Chem. 2009, 19, 8902. (28) Kumar, A.; Whitesides, G. M. Appl. Phys. Lett. 1993, 63, 2002. (29) Ruiz, S. A.; Chen, C. S. Soft Matter 2007, 3, 168. (30) Perl, A.; Reinhoudt, D. N.; Huskens, J. Adv. Mater. 2009, 21, 2257. (31) Bernard, A.; Delamarche, E.; Schmid, H.; Michel, B.; Bosshard, R. H.; Biebuyck, H. Langmuir 1998, 14, 2225. (32) Ludden, M. J. W.; Mulder, A.; Schulze, K.; Subramaniam, V.; Tampe, R.; Huskens, J. Chem.;Eur. J. 2008, 14, 2044. (33) Lange, S. A.; Benes, V.; Kern, D. P.; H€ober, J. K. H.; Bernard, A. Anal. Chem. 2004, 76, 1641. (34) Thibault, C.; Le Berre, V.; Casimirius, S.; Trevisol, E.; Franc-ois, J.; Vieu, C. J. Nanobiotechnol. 2005, 3, 7. (35) Rozkiewicz, D. I.; Gierlich, J.; Burley, G. A.; Gutsmiedl, K.; Carell, T.; Ravoo, B. J.; Reinhoudt, D. N. ChemBioChem 2007, 8, 1997. (36) Rozkiewicz, D. I.; Brugman, W.; Kerkhoven, R. M.; Ravoo, B. J.; Reinhoudt, D. N. J. Am. Chem. Soc. 2007, 129, 11593. (37) Michel, O.; Ravoo, B. J. Langmuir 2008, 24, 12116. (38) Godula, K.; Rabuka, D.; Nam, K. T.; Bertozzi, C. R. Angew. Chem., Int. Ed. 2009, 48, 4973.
4934 DOI: 10.1021/la903569v
Wendeln et al.
microspotting are high edge resolution (better than 100 nm), reproducible probe density, homogeneous spots, and short reaction times.
Experimental Section General. Chemicals were purchased from Sigma Aldrich or from Acros Organics and used without further purification, unless otherwise noted. 11-Bromoundecyltrichlorosilane was obtained from ABCR. Tetrahydrofuran was dried by distillation from potassium/benzophenone. Dichloromethane was distilled from CaH2. Methanol and dimethylformamide (DMF) were dried by using molecular sieves (3 A˚) for at least 3 days. Reactions were monitored by thin-layer chromatography (TLC) where possible, which was performed with 0.2 mm Merck precoated silica gel 60 F254 aluminum sheets. For column chromatography, silica gel 60 (0.063-0.2 mm, Merck) was used. NMR spectra were recorded on Bruker spectrometers (AV 300, AV 400) by using deuterated chloroform or deuterated methanol as solvents. Chemical shifts are given in ppm and expressed relative to the carbon or hydrogen signals of the solvents. Mass spectra were recorded by using a MicroTof spectrometer (Bruker). Silicon wafers (B-doped, 1-0-0 orientation, 20-30 Ω) were kindly donated by Siltronic AG. Substrates were cleaned with dichloromethane p.a., dry ethanol, and Milli-Q water, which was prepared from distilled water using a PureLab UHQ deionization system (Elga). Fluorescein isothiocyanate (FITC)-labeled concanavalin A (ConA) was purchased from Sigma Aldrich. Tetramethylrhodamine isothiocyanate (TRITC)-labeled peanut agglutinin (PNA) was obtained from Vector Laboratories. 3,6-Endoxo-Δ4-tetrahydrophthalimide was synthesized according to a literature procedure.39 Detailed information concerning the synthesis of compounds 1-7 is provided in the Supporting Information. SAM Preparation. Maleimide SAMs were prepared as reported by Goedel et al.40 with small modifications. Glass slides and Si-wafers were cut into pieces of 1.6 2.6 cm2, cleaned by sonication in pentane, acetone, and Milli-Q water, dried, and then immersed into a freshly prepared piranha solution (H2O2/H2SO4 = 1/3) for 30 min. The substrates were washed with Milli-Q water, dried, and put in a stirred solution of 11-bromoundecyltrichlorosilane in toluene (0.1 vol %) for 30 min. The bromo-terminated SAMs were cleaned by washing with ethanol, dried, and immersed into a solution of 3,6-endoxo-Δ4-tetrahydrophthalimide (20 mM) in K2CO3 saturated, dry DMF. The surface reaction was carried out for 16 h at 65 °C under an atmosphere of argon. After washing the surfaces with Milli-Q water, dichloromethane, and ethanol, the substrates were heated in an oven for 2 h at 170 °C to transform the protected maleimide-terminated SAMs into the maleimideterminated SAMs. PDMS Stamps. Poly(dimethylsiloxane) (PDMS) stamps were prepared by casting a 10:1 (v/v) mixture of PDMS and curing agent (Sylgard 184, Dow Corning) on a patterned silicon master. The PDMS was kept in an oven at 80 °C overnight for curing. The patterned compartments were cut out with a knife and oxidized in a UV-ozonizer (PSD-UV, Novascan Technologies Inc.) for 30 min prior to their use. Flat PDMS stamps were made analogously, using a flat silicon master. Microcontact Printing. The surface of the freshly oxidized PDMS stamp was covered by a few drops of the ethanolic ink solutions (2 mM for the carbohydrate-cyclopentadiene conjugates and the dansyl derivate, 10 mM for the galactose-furan conjugate). After 1 min of incubation, the stamps were blow dried and then carefully placed on the maleimide-terminated SAMs. The printing time varied from 5 to 15 min for the cyclopentadiene inks and from 30 min to 3 h for the furan ink. If printing was not (39) Zhu, J.; Ganton, M. D.; Kerr, M. A.; Worketin, M. S. J. Am. Chem. Soc. 2007, 129, 4904. (40) Wang, Y.; Cai, J.; Rauscher, H.; Behm, R. J.; Goedel, W. A. Chem.;Eur. J. 2005, 11, 3968.
Langmuir 2010, 26(7), 4933–4940
Wendeln et al. carried out at room temperature, the substrates were placed together with the stamp in a previously tempered oven. After printing the stamps were removed from the surfaces and the substrates were washed with dichloromethane and ethanol, sonicated in ethanol, and dried. Further printing steps on the obtained carbohydrate arrays were carried out in the same way. Lectin-Carbohydrate Interactions. In order to inactivate remaining maleimide groups of the arrays, the substrates were immersed into a 10 mM solution of L-cysteine in PBS buffer (pH = 7.5) for 1 h, washed with Milli-Q water, and dried. The surfaces of the carbohydrate arrays were covered by a solution of 1 μg of fluorescence-labeled lectin (FITC-labeled Con A or TRITC-labeled PNA) in 100 μL of HEPES buffer (20 mM HEPES, pH = 7.5, 0.15 M NaCl, 1.0 mM CaCl2). In the case of ConA, MnCl2 was added to a concentration of 1 mM. After 90 min, the arrays were washed with the same buffer, rinsed with Milli-Q water, dried, and analyzed. To reduce nonspecific protein adsorption, the arrays were incubated with a 3% bovine serum albumin (BSA) solution in PBS buffer (pH = 7.5, 0.1% Tween 20) for 30 min prior to lectin screening and afterward washed in the same buffer without BSA (3 10 min). Contact Angle Measurements. Water contact angles were measured by means of the sessile drop method on a DSA 100 goniometer (Kr€ uss GmbH Wissenschaftliche Laborger€ate). Advancing and receding contact angles were measured on silicon substrates, and at least three measurements were performed for every sample. The evaluation of the measurements was done using the software Drop Shape Analysis. X-ray Photoelectron Spectroscopy. X-ray photoelectron spectra were recorded with the Kratos Axis Ultra system (Kratos Analytical) and monochromatized AlKa radiation (1486.6 eV) as the excitation source. The obtained spectra were evaluated by using the software Kratos Vision 2 and were referenced by setting the C(1s)-peak of the saturated hydrocarbons to 284.5 eV. All measurements were carried out on glass substrates. Time-of-Flight Secondary Ion Mass Spectrometry. SIMS experiments were performed using a type IV compatible TOFSIMS instrument equipped with a liquid metal ion gun (IONTOF GmbH). As primary ions Bi3þ with an energy of 25 keV were used. The samples were imaged with a primary ion dose density of up to 5 1012 ions/cm2. All measurements were carried out on silicon substrates. Fluorescence Microscopy. Fluorescence microscopy images were made by using an Olympus inverted research microscope CKX41 equipped with a mercury burner U-RFL-T as light source and a DX 20 L-FW camera (Kappa opto-electronics GmbH) for image acquisition. The camera was controlled by the program Kappa CameraControl (version 2.7.5.7032). All investigations were carried out on glass substrates.
Results and Discussion The carbohydrate inks used to print carbohydrate microarrays are shown in Chart 1. For a covalent immobilization through a Diels-Alder reaction, it was necessary to synthesize diene-functionalized carbohydrates which would allow a stable linkage with a dienophile-terminated SAM. To guarantee unhindered carbohydrate-protein interactions, the carbohydrate and the diene were separated by a spacer. For this purpose, oligo(ethylene glycol) chains are commonly used, as it is known that such a spacer is conformationally flexible, water-soluble, and resistant to protein adsorption.41 Using oligo(ethylene glycol) as the spacer, we derivatized different carbohydrates by connecting them to cyclopentadiene, leading to compounds 1-5 shown in Chart 1. The β-D-galactoside 1, R-D-mannoside 2, (41) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, 10714.
Langmuir 2010, 26(7), 4933–4940
Article Chart 1. Molecular Structures of Carbohydrate inks 1-5 and 7 and Fluorescent Ink 6
and β-D-glucoside 3, and R-D-mannoside 3 were synthesized as reported by Houseman and Mrksich, with small modifications in the synthetic procedure. β-D-lactoside 4 and β-D-maltoside 5 were synthesized analogously. In order to verify the printing of cyclopentadiene conjugates on maleimide surfaces in general, the dansyl group was connected to the same cyclopentadieneterminated linker, leading to compound 6. The fluorescent dansyl group allowed the detection of printed molecules on the surface by fluorescence microscopy. Finally, we also synthesized the β-D-galactoside 7, in which galactose is coupled with a furan ring. Furan is much less reactive toward dienophiles in comparison with cyclopentadiene. From this perspective, this compound is also of interest to test the efficiency of μCP of carbohydrate inks to print microarrrays. The synthesis of 1-5 and 7 was carried out using trichloroimidate glycosidation chemistry. Details of the synthesis are reported in the Supporting Information. The spectroscopic and analytical data for 1-7 are consistent with their molecular structure. A dienophile-functionalized surface was required for an effective immobilization of the carbohydrate inks. We selected the maleimide group as the dienophile because maleimides are stable yet highly reactive toward dienes and maleimide-terminated SAMs are relatively easy to synthesize. The most common approach to form maleimide-terminated SAMs is based on treating an amino monolayer with a heterobifunctional crosslinker that carries a maleimide group on one side and an active ester at the other side. The active ester reacts with the aminoterminated surface, forming a stable amide bond and leading to the generation of a maleimide covered substrate. However, we preferred another established procedure (Figure 1).40 First, activated glass or silicon substrates are treated with 11-bromoundecyltrichlorosilane in toluene to generate a bromo-terminated SAM (Figure 1A). The bromo-terminated SAM is then converted into a 3,6-endoxo-Δ4-tetrahydrophthalimide-terminated SAM by nucleophilic substitution of the bromine by 3,6-endoxo-Δ4-tetrahydrophthalimide39 in K2CO3 saturated DMF (Figure 1B). Heating in an oven at 170 °C leads to the cleavage of the furan ring in a DOI: 10.1021/la903569v
4935
Article
Wendeln et al.
Figure 1. Surface reaction sequence for the preparation of the maleimide-terminated monolayer (A-C) and C(1s) signals in the XPS spectra of the bromo-terminated SAM (D), the endoxotetrahydrophthalimide-terminated monolayer (E), and the maleimide-terminated monolayer (F).
Figure 2. Schematic overview of the immobilization of a carbohydrate-cyclopentadiene conjugate on a maleimide-terminated SAM induced by μCP (“microcontact chemistry”).
retro-Diels-Alder reaction and the formation of the maleimideterminated SAM (Figure 1C). The consecutive reactions of the SAMs could be verified by analyzing the C(1s) signals in the X-ray photoelectron spectroscopy (XPS) spectra (see Figure 1). Whereas the C(1s) signal of the bromo-terminated SAM (Figure 1D) shows the presence of carbon atoms in C-Br and C-C/C-H bonds, the XPS spectrum of the endoxotetrahydrophthalimide-terminated SAM (Figure 1E) shows both NCdO and C-O carbon atoms, confirming the replacement of the bromine by endoxotetra4936 DOI: 10.1021/la903569v
hydrophthalimide. The thermal elimination of the furan by retro-Diels-Alder reaction leads to the disappearance of the C-O shoulder in the C(1s) signal (Figure 1F) and therefore proves the successful preparation of the maleimide-terminated SAM. The Diels-Alder reaction by μCP was tested by printing the carbohydrate-cyclopentadiene conjugates (1-5) on maleimide SAMs (Figure 2). To this end, the surface of oxidized, patterned PDMS stamps were covered with 2 mM ethanolic solutions of the carbohydrates and blow dried after 1 min. Subsequently, the Langmuir 2010, 26(7), 4933–4940
Wendeln et al.
Article
Figure 3. Microscopy image of water droplets selectively condensed in the areas where galactose-cyclopentadiene conjugate 1 was immobilized by μCP on a maleimide-terminated SAM.
Figure 4. Microscopy image of water droplets selectively condensed in the areas where galactose-furan conjugate 7 was immobilized by μCP at 80 °C on a maleimide-terminated SAM.
stamps were placed on the maleimide-terminated SAMs on glass substrates to allow chemical reaction in the contact area. The stamps were removed after a reaction time of 5-15 min. The presence of the carbohydrates on the surfaces was investigated by several methods. First, a significant change of the surface hydrophilicity was detected, clearly indicating the immobilization of the polar carbohydrates on the substrate. It was possible to visualize the change in surface hydrophilicity by condensing water selectively in the carbohydrate covered areas. Figure 3 shows a maleimide covered glass slide under the optical microscope on which the galactoside 1 was printed with a structured stamp (5 μm lines that are spaced by a 15 μm gap) and water was condensed by breathing onto the surface. It is obvious that the water droplets exclusively form in the regions where the hydrophilic galactoside 1 was immobilized. Also, the elongated shape of the drops reflects the line pattern of the stamp. The observed change in hydrophilicity was quantified by contact angle measurements. To this end, the entire surface was reacted homogeneously with the carbohydrates by using a flat PDMS stamp for μCP. After a printing time of 15 min, the water contact angle decreased from 82°/62° (adv/rec) for the maleimide SAM to 48°/15° for the carbohydrate SAMs. Interestingly, if the printing time was only 5 min, almost the same contact angles were obtained, indicating a fast surface reaction by μCP. We emphasize that Diels-Alder reactions of an immobilized dienophile with cyclopentadiene from solution typically requires 2-12 h reaction time!8,22 We also investigated the printing of the furan-galactose conjugate 7 on maleimide-terminated SAMs. Although the Diels-Alder reaction of furan with maleimide is much slower42 than the corresponding reaction with cyclopentadiene, it is conceivable that μCP could accelerate the reaction as it is solvent free and high concentrations of molecules are present on the stamp surface.27 A flat, oxidized PDMS stamp was inked with a 10 mM solution of galactoside 7 and printed for 1 h at room temperature. Contact angle measurement indicated a modest decrease of 82°/62° (adv/rec) for the maleimide SAM to 75°/31° for the printed SAM. This result shows that a reaction took place, but the small decrease of the contact angles suggests the surface coverage of the furan-galactose conjugate 7 is inferior to that of the cyclopentadiene-galactose conjugate 1 printed under the
same conditions. In addition, also the large hysteresis (>40°) indicates a very heterogeneous and less dense surface coating. To improve the surface reaction of galactoside 7, μCP was carried out at higher temperatures. μCP at 60 °C for 3 h led to a decrease of the contact angles to 50°/12°, and μCP at 80 °C for 1 h resulted in contact angles of 44°/12°, signaling a faster reaction and a much higher surface coverage of carbohydrate at this temperature. Longer printing times at 80 °C did not result in a noticeable further decrease of the contact angles. If galactoside 7 was printed with a patterned stamp (5 μm lines spaced by 15 μm) for 1 h at 80 °C and water was condensed as previously described by breathing onto the surface, the patterned immobilization of the galactoside could be easily identified by condensation of water on the surface (Figure 4). The shape and orientation of the water drops are in accordance with the dimension of the stamp. Although the results described above demonstrate that μCP of furan-carbohydrate conjugates on maleimide-terminated SAMs via Diels-Alder reaction is possible, this immobilization method does not fulfill the requirements for carbohydrate microarray fabrication, since in this case μCP requires high temperatures and long reaction times. This could result in contamination of the substrates by PDMS residues, which would lead to false hits in protein screening. In addition, at elevated temperatures, the PDMS stamp is sensitive to deformations, leading to patterns that show defects and have a low resolution. Therefore, we concentrated on the investigation of cyclopentadiene-functionalized carbohydrates, as these molecules can be printed very effectively at room temperature. Besides the change in surface hydrophilicity, the immobilization of the carbohydrate-cyclopentadiene conjugates can also be proven by XPS.43 Figure 5 shows the XPS signals in the C(1s) region of the maleimide-terminated surface (Figure 5A) and the C(1s) regions of two “carbohydrate chips”, which were made by printing galactoside 1 (Figure 5B) and maltoside 5 (Figure 5C), respectively, on a maleimide-terminated SAM by using a flat PDMS stamp. In the spectrum of the maleimide covered substrate, a peak at 285 eV indicates the carbon atoms of the alkyl chains and a peak around 288 eV indicates the NCO groups of the maleimide rings. The carbohydrate covered surfaces show the same signals, but in addition characteristic peaks around 286 eV point at the C-O single bonds of the carbohydrate-cyclopentadiene conjugates. This result
^ (42) Ruli^sek, L.; Sebek, P.; Havlas, Z.; Hrabal, R.; Capek, P.; Svatos, A. J. Org. Chem. 2005, 70, 6295.
Langmuir 2010, 26(7), 4933–4940
(43) Dhayal, M.; Ratner, D. M. Langmuir 2009, 25, 2181.
DOI: 10.1021/la903569v
4937
Article
Wendeln et al.
Figure 5. Comparison of the C(1s) signals in the XPS spectra of (A) a maleimide-terminated SAM, (B) a galactoside covered SAM made by μCP of galactose-cyclopentadiene conjugate 1, and (C) a maltoside covered SAM made by μCP of maltose-cyclopentadiene conjugate 5.
Figure 7. TOF-SIMS analysis of an array of maltoside 5. The carbohydrate was printed with an oxidized PDMS stamp (5 μm lines spaced by 20 μm). The measurements were carried out in the negative ion mode, and Bi3þ was used as primary ion. The field of view is 40 40 μm2.
Figure 6. Fluorescence microscopy image of (A) dansylcyclopentadiene derivate 6 printed in a line pattern on a maleimide SAM on a glass slide and (B) mannoside 2 printed in a dot pattern and all remaining free space filled by printing dansylcyclopentadiene derivate 6 with a flat stamp.
clearly indicates the immobilization of the carbohydrates on the surface. Compared with the galactose chip, the C-O peak in the spectrum of the maltose chip is more distinct, which is in accordance with the higher proportion of C-O bonds on the surface. Moreover, the intensity of the peaks at 288 eV increased slightly due to the presence of anomeric carbohydrate carbon atoms. Another experiment that proves the immobilization of cyclopentadiene conjugates onto maleimide SAMs by μCP is shown in Figure 6. The dansyl derivate 6 was printed with a patterned 4938 DOI: 10.1021/la903569v
PDMS stamp onto a maleimide-terminated glass slide (Figure 6A). The immobilization of the fluorophore in the contact areas was visualized with fluorescence microscopy. The pattern is consistent with the structure of the stamp (5 μm lines spaced by 15 μm). Even after intensive washing with dichloromethane and ethanol, the pattern did not vanish, indicating the covalent binding of the molecule onto the surface. It is also possible to visualize the immobilization of the carbohydrates in an indirect way by using the dansyl-cyclopentadienyl conjugate: first, mannoside 2 was printed (15 min) on a maleimide-terminated SAM on a glass slide with a stamp that had a dotted microstructure; second, the remaining areas were filled by printing dansyl derivate 6 with a flat stamp (15 min). Fluorescence microscopy clearly reveals the negative image of the dotted pattern of the stamp (Figure 6B). Two important conclusions can be drawn from this experiment. First of all, μCP of the carbohydrate-cyclopentadiene conjugates proceeds within 15 min and results in a high surface coverage, since no significant immobilization of the dansyl derivate is possible in the carbohydrate coated areas. Second, μCP of cyclopentadienyl derivates is possible with rather high accuracy and good resolution. We also investigated our carbohydrate chips by time-of-flight secondary ion mass spectrometry (TOF-SIMS).44 Figure 7 shows the analysis of an array, which was made by printing maltoside 5 on a maleimide-terminated monolayer. The measurements were carried out in the negative ion mode, and each image visualizes the generation of specific, negatively charged secondary ions depending on the surface location. In the areas where the carbohydrates were printed, the formation of characteristic, oxygen rich anions was obtained, namely, C3H3O2-, C2H3O2-, C2H2O2-, CHO2-, and C2HO-. This result is in accordance with the oxygen rich molecular structure of the carbohydrate-cyclopentadiene conjugates. The contrast in the images of these anions is high, except for the CHO2- and for the C2HO- ion, which shows that these fragments can also be generated by the fragmentation of the maleimide group, but with less probability compared to the surface (44) Arlinghaus, H. F. Appl. Surf. Sci. 2008, 255, 1058.
Langmuir 2010, 26(7), 4933–4940
Wendeln et al.
Figure 8. TOF-SIMS analysis of an array, which was made by μCP of maltoside 5 (near vertical lines), followed by μCP of lactoside 4 in the perpendicular direction (near horizontal lines). The field of view is 100 100 μm2. The measurement was carried out in the negative ion mode.
linked carbohydrates. Among the oxygen rich fragments also the nitrogen containing anion CNO- attests the immobilization of the carbohydrate-cyclopentadiene conjugate on the surface, since this ion is especially formed within the carbohydrate coated areas, although the image indicates that it can also be generated by the maleimide group. Also, the patterns of the C2H- and the SiO3- anion indicate the successful immobilization of the carbohydrates. In the case of SiO3-, the fragment is not due to the maleimide-terminated monolayer or to the carbohydrate conjugates, but it has its origin in the underlying SiO2 layer. It can be assumed that the organic monolayer is significantly thicker in the areas where the carbohydrates were printed, which hinders desorption of the SiO3- anion by covering the SiO2 substrate. We also investigated a carbohydrate microarray that consists of two carbohydrates by TOF-SIMS. Figure 8 shows the analysis of an array, which was made by consecutive μCP of maltoside 5 with a patterned stamp (5 μm lines spaced by 25 μm, near vertical lines) and lactoside 4 perpendicularly with a second patterned stamp (5 μm lines spaced by 20 μm, near horizontal lines). Both carbohydrates decompose with the formation of characteristic secondary ions under Bi3þ bombardment, as discussed above. The observed lattice structure is in accordance with the dimensions of the used stamps, and this again shows that μCP of carbohydrate-cyclopentadiene conjugates is possible with good resolution. The same array was also analyzed in the positive ion mode. Figure 9 shows the results of the measurements. The field of view is 50 50 μm2 and therefore represents a zoom-in compared to Figure 8. In accordance with discussion above, oxygen rich fragments such as the C4H5O2þ, the C2H5Oþ, and the CH2Oþ cations indicate the immobilization of the carbohydrates. These cations show high contrast and therefore seem to be characteristic for the carbohydrate conjugates. On the other hand, the C3H5þ fragment is an example of a fragment that did lead to an image with low contrast as it can be easily formed from alkyl chains which are present in the maleimide-termiated monolayer as well as in the carbohydrate conjugates. Also, the CH4Nþ fragment Langmuir 2010, 26(7), 4933–4940
Article
Figure 9. TOF-SIMS analysis of an array of maltoside 5 (vertical lines) and lactoside 4 (horizontal lines) in the positive ion mode. The field of view is 50 50 μm2.
indicates the presence of the carbohydrates in the expected regions. An interesting ion that shows the carbohydrate pattern is the C7H7þ cation. This cation is prominent because of its high stability and can be generated from unsaturated hydrocarbon chains. The formation of the cation indicates the presence of the cyclopentadiene in the carbohydrate conjugates on the surface because this cyclic system can easily lead to the C7H7þ cation. Nevertheless, the presence of the C7H7þ cation cannot be taken as a direct proof for the covalent binding by Diels-Alder reaction. TOF-SIMS indicated the immobilization of carbohydrates on the surface also in an indirect way, as Naþ cations (see Figure 9), Kþ cations, and also Cl- and Br- anions (data not shown) could preferentially be found on the carbohydrate coated areas. This can be explained from the accumulation of salt traces on the more hydrophilic carbohydrate containing regions from washing procedures, which is consistent with the contact angle measurements. The detection of silicon cations preferably in the carbohydrate free regions verifies the presence of the carbohydrates on the surface in the same way as discussed for the SiO3- anion. In addition, we studied the binding of two lectins to the carbohydrate arrays prepared by μCP (Figure 10). Lactoside 4 was printed on a maleimide-terminated glass slide, and remaining maleimide groups were passivated by reacting them with cysteine. The resulting “carbohydrate chip” was then exposed to a solution of TRITC-labeled PNA in HEPES buffer. After washing the chip in buffer solution, the array was analyzed with fluorescence microscopy (Figure 10A). The lectin is patterned in fluorescent stripes, which match the pattern of the stamp (5 μm lines spaced 20 μm apart). This observation clearly indicates that PNA binds selectively to lactoside, which is in line with the literature.45,46 This observation also indicates that the density of lactoside on the surface must be substantial, otherwise PNA cannot bind in a multivalent fashion (i.e., resistant to extensive washing). If the same experiment is carried out without a carbohydrate ink (pure ethanol is used as the ink solution instead), no fluorescent pattern can be found. This observation eliminates the possibility that the (45) Dam, T. K.; Brewer, C. F. Chem. Rev. 2002, 102, 387. (46) Lis, H.; Sharon, N. Chem. Rev. 1998, 98, 637.
DOI: 10.1021/la903569v
4939
Article
Wendeln et al.
that the horizontal stripes are interrupted by 5 μm gaps at a spacing of 25 μm. These values fit to the dimension of the PDMS stamp that was used to print glucoside 3. In addition, no (or at least, very little) galactoside 1 was immobilized in the areas which were reacted with the glucoside in the first printing step, since no PNA binding was observed and clear-cut gaps were obtained. These results are in line with the μCP of the dansylcyclopentadiene conjugate 6 shown in Figure 6. Using μCP, we also prepared an array that consists of three carbohydrates. First, mannoside 2 was printed in horizontal lines. Second, maltoside 5 was printed in vertical lines. Third, the remaining free areas were filled with glucoside 3 by printing with a flat stamp. Exposure of this array with FITC-ConA resulted in a lattice structure (Figure 10D), showing that ConA binds to mannoside (near horizontal lines, 5 μm spaced by 25 μm) as well as to maltoside (near vertical lines, 5 μm spaced by 20 μm) but not to glucoside (20 25 μm2 rectangles interspace). In order to reduce the background fluorescence of the microarrays, a BSA blocking step can be introduced. This leads to a decrease of nonspecific lectin adsorption and therefore increases the contrast of the chips, as can be seen from Figure 10E and F. The images show the results of experiments analogous to those shown in Figure 10A (lactoside 4 printed in 5 μm lines spaced by 15 μm) and B (mannoside 2 printed in 5 μm lines and 20 μm interspace filled with galactoside 1). Figure 10. Fluorescence microscopy images of TRITC-labeled PNA (A, C, and E) and FITC-labeled ConA (B, D, and F) on carbohydrate microarrays made by μCP. Arrays were made of (A) a single carbohydrate (lactoside 4), (B) two carbohydrates (galactoside 1 and mannoside 2), (C) two carbohydrates (galactoside 1 and glucoside 3), or (D) three carbohydrates (mannoside 2, glucoside 3, maltoside 5). (E) and (F) show fluorescence images of experiments similar to those described in (A) and (B) with an additional BSA blocking step to reduce nonspecific lectin adsorption. See text for details.
lectin simply adsorbed to PDMS residues, which could have been transferred onto the surface during μCP. Beside the preparation of arrays of a single carbohydrate, chips with two different carbohydrates were also made. This was done, for example, by first printing mannoside 2 on a maleimide glass slide with a patterned stamp and then filling the free space with galactoside 1 by printing with a flat stamp. The array was exposed to a solution of fluorescein-labeled ConA in HEPES buffer, washed, and analyzed. It was found that ConA only binds to the immobilized mannoside but not to galactoside, resulting in a fluorescent pattern which is identical to the pattern of the used stamp (Figure 10B). Another array consisting of two carbohydrates was made by first printing glucoside 3 in vertical lines on a maleimide substrate and then printing galactoside 1 in the perpendicular direction (15 min printing time each). After blocking the free maleimide groups with cysteine, the array was exposed to a solution of PNA. The result of the fluorescence imaging can be seen in Figure 10C. The fluorescent PNA binds to galactoside but not to glucoside, so
4940 DOI: 10.1021/la903569v
Conclusion In conclusion, we have demonstrated the successful μCP of cyclopentadiene and furan derivates of carbohydrates on maleimide-terminated SAMs with a range of surface analytical methods. μCP allows the immobilization of carbohydrates on surfaces in high density and with high resolution. With μCP, Diels-Alder reactions on surfaces can be conducted in minutes, not hours. Using this method, carbohydrate microarrays can be produced, showing the applicability of this technique for the immobilization of biomolecules on surfaces. The Diels-Alder reaction between cyclopentadiene and maleimide is particularly suited for the production of biological arrays, since both diene and dienophile are orthogonal to most functional groups that are commonly found in nature. We contend that the combination of μCP and Diels-Alder reactions is an excellent tool for the production of microarrays and might be of great value for proteomics, genomics, and glycomics. Acknowledgment. Silicon wafers were kindly donated by Siltronic AG. XPS measurements were carried out by Dr. Matthias Rinke and Dr. Uta Rodehorst. The Deutsche Forschungsgemeinschaft is acknowledged for financial support of this work (Grant Ra 1732/2-1). Supporting Information Available: Synthesis and analysis of compounds 1-7. This material is available free of charge via the Internet at http://pubs.acs.org.
Langmuir 2010, 26(7), 4933–4940
