Synthesis and Characterization of a Photocleavable Cross-Linker and

Rong Wang , Funing Yan , Dengli Qiu , Jae-Sun Jeong , Qiaoling Jin , Tae-Young Kim ... Guangchang Zhou , Xiaoliang Yan , Ding Wu , and Stephen J. Kron...
0 downloads 0 Views 225KB Size
1030

Bioconjugate Chem. 2004, 15, 1030−1036

Synthesis and Characterization of a Photocleavable Cross-Linker and Its Application on Tunable Surface Modification and Protein Photodelivery Funing Yan,†,§ Liaohai Chen,*,§ Qiling Tang,† and Rong Wang*,† Illinois Institute of Technology, Chicago, Illinois 60616, and Argonne National Laboratory, Argonne, Illinois 60439. Received April 21, 2004; Revised Manuscript Received June 29, 2004

A heterobifunctional photocleavable cross-linker based on an o-nitrobenzyl ester moiety was synthesized. The cross-linker has N-hydroxysuccinimidyl and disulfide groups attached at each end and thus can anchor a protein to a gold-coated substrate surface. Steady-state spectroscopic studies suggest that the cross-linker undergoes a clean C-O fragmentation upon irradiation with a quantum yield of 0.1. Consequently, immobilized proteins (such as avidin or antibodies) on a substrate surface can be released efficiently (>95%) under UV irradiation (λ > 300 nm) without degrading the protein functionality. We also demonstrated protein delivery via bioconjugation of protein molecules to a goldcoated atomic-force microscope (AFM) tip. When the proteins are photoreleased from the AFM tip, they are delivered to the substrate surface as protein clusters of uniform size. This has been confirmed using both AFM and fluorescence microscopy. The application of bioconjugation in this study opens a new avenue for tunable surface modification and controllable protein delivery in studies of biological systems on the nanometer scale.

INTRODUCTION

Substrate surface modification with proteins is essential in bioassaying, bioengineering, and biotechnology (1). While many efforts are focused on the covalent linkage of proteins on a substrate surface (2, 3), it is important to develop methods for detaching the anchored proteins from the substrate surface as desired. One of the practices commonly used to achieve this goal is the use of a reactive enzyme (4) to cleave the bond that holds the protein to the substrate surface. This approach may be cumbersome because it involves multiple steps and requires accommodation of enzyme working conditions. An alternative is to anchor proteins to a substrate surface using a photocleavable bond (5), so that the proteins can be detached by a photofragmentation reaction. Unlike the enzyme method, the photodetachment of proteins is clean and easy to carry out. No reagents are needed. More importantly, since light can be focused to a small spot (100-400 nm in diameter, depending on the diffraction limitation), the photofragmentation reaction can be achieved in a confined area, allowing detachment of proteins at a precise local position. Immobilization of proteins using a heterobifunctional and photoactivatable cross-linker will open new avenues for manipulating the attachment and detachment of biomolecules to a surface. Among the photocleavable moieties used to cross-link proteins to a substrate surface, the o-nitrobenzyl group is popular due to unique properties that include stability under ambient light, clean cleavage upon exposure to UV irradiation, and fast fragmentation reactions (within nanoseconds) upon photoexcitation (6, 7). This group has been used in myriad solid-phase synthesis, protein isolation, and protein purification experiments (8-12). In this * Address correspondence to these authors. e-mail: lhchen@ anl.gov or [email protected]. † Illinois Institute of Technology. § Argonne National Laboratory.

paper, we report the development of a heterobifunctional photocleavable cross-linker that uses the o-nitrobenzyl ester moiety as a core. One terminal of the cross-linker, N-hydroxysuccinimidyl, allows a protein to be conjugated. The other terminal, a disulfide group, allows chemical bonding to a gold-coated substrate. The clean C-O photofragmentation was confirmed by our steady-state spectroscopic studies. Proteins (such as avidin or antibodies) anchored to a gold substrate surface by the crosslinker can be efficiently (>95%) photodetached under UV irradiation without degrading protein functions. We also demonstrated that proteins anchored to the AFM tip can be photodelivered to a substrate surface. As a complement to other protein delivery methods such as Dip-pen technology (13, 14), the method developed in this paper allows for precise and controllable protein delivery. EXPERIMENTAL SECTION

Materials. Reagent grade solvents were used without any purification unless specified; chemicals including 5-hydroxy-2-nitrobenzyl alcohol, 3,4-dihydro-2H-pyran, p-toluenesulfonic acid, N,N′-dicyclohexylcarbodiimide (DCC), thioctic acid, succinic anhydride, dimethyl aminopyridine (DMAP), N-hydroxy succinimide, and general organic solvents were purchased from the Sigma-Aldrich Chemical Co. (St. Louis, MO). THF was distilled before use. N-Succinimidyl 3-(2-pyridyldithio)propionate (SPDP) and Alexa Fluoro 488 biocytin were purchased from Molecular Probes, Inc. (Benertin, OR). Synthesis of Photolabile Cross-Linker. Synthesis of 4-Nitro-3-(tetrahydropyran-2-yloxymethyl)phenol (2). Nitrobenzyl alcohol (3.4 g, 0.02 mol) and p-toluenesulfonic acid (0.02 g, 0.1 mmol) were dissolved in 30 mL of dried THF. Under N2 protection, 2.7 mL of 3,4-dihydro-2Hpyran was added dropwise. After the reaction was refluxed for 10 h, THF was evaporated under vacuum. The residue was chromatographed on silica gel using ethyl acetate/n-hexane (1:4) and yielded a white crystal-

10.1021/bc049901d CCC: $27.50 © 2004 American Chemical Society Published on Web 08/24/2004

Photocleavable Cross-Linker

line solid. Recrystallization in ethyl acetate/n-hexane (1: 1) resulted in 4.3 g of 2 (yield 86%). Mp 141-143 °C, 1H NMR (300 MHz, CDCl3): δ 8.13 (d, 1H), 7.36 (d, 1H), 6.79 (d, 1H), 5.18 (m, 2H), 4.80 (t, 1H), 3.98 (m, 1H), 3.60 (m, 1H), 1.90 (m, 2H), 1.60 (m, 4H). Synthesis of 5-[1,2]-Dithiolan-3-ylpentanoic Acid 4-Nitro-3-(tetrahydropyran-2-yloxymethyl)phenyl Ester (3). 2 (1.0 g, 4 mmol), thioctic acid (1.65 g, 8 mmol), and DCC (1.65 g, 8 mmol) were dissolved in 20 mL of dried THF. After the reaction was refluxed under N2 protection for 12 h, the resulting suspension was filtered out and the THF was then evaporated under vacuum. The residue was chromatographed on silica gel using dichloromethane/ n-hexane (5:1) to give 1.5 g of 3 (yield 85%). Yellow gel, 1 H NMR (300 MHz, CDCl3): δ 8.16 (d, 1H), 7.53 (d, 1H), 7.18 (d, 1H), 5.00 (m, 2H), 4.76 (t, 1H), 3.85 (m, 1H), 3.57 (m, 2H), 3.11 (m, 2H), 2.60 (t, 2H), 2.42 (m, 1H), 1.92 (m, 1H), 1.90-1.50 (m, 12H). Synthesis of 5-Thioctic Acid 3-Hydroxymethyl-4-nitrophenyl Ester (4) 3. (2.2 g, 5 mmol) was dissolved in dilute hydrochloride H2O/THF solution (0.2 M, 1:2 V/V) and stirred overnight at room temperature. Then the mixture was poured into ethyl acetate followed by washing three times with 1% aqueous NaHCO3 solution. After normal workup, the organic phase was chromatographed on silica gel using dichloromethane to give 1.6 g of 4 (yield 90%). Yellow gel, 1H NMR (300 MHz, CDCl3): δ 8.20 (d, 1H), 7.58 (d, 1H), 7.20 (d, 1H), 5.01 (s, 2H), 3.60 (m, 1H), 3.18 (m, 2H), 2.61 (t, 2H), 2.40 (m, 1H), 1.78 (m, 4H), 1.58 (t, 2H). Synthesis of Succinic Acid Mono-[5-(5-[1,2]dithiolan3-yl-pentanoyloxy)-2-nitrobenzyl] Ester (5). 4 (0.7 g, 2 mmol), succinic anhydride (0.4 g, 4 mmol), and 4-dimethylaminopyridine (DMAP) (0.12 g, 1 mmol) were dissolved in 15 mL of dichloromethane and refluxed under N2 protection for 10 h. The mixture was washed three times by distilled water followed by normal workup and chromatographed on silica gel using dichloromethane/ methanol (9:1) to give 0.7 g of 5 (yield 75%). Yellow gel, 1 H NMR (300 MHz, CDCl3): δ 8.20 (d, 1H), 7.39 (d, 1H), 7.21 (d, 1H), 5.59 (s, 2H), 3.60 (m, 1H), 3.18 (m, 2H), 2.72 (t, 2H), 2.61 (t, 2H), 2.45 (m, 1H), 1.97 (m, 1H), 1.78 (m, 4H), 1.58 (t, 2H). Synthesis of Succinic Acid Succinimidyl Ester 5-Thioyloxy-2-nitrobenzyl Ester (SSTN) (6). 5 (0.9 g, 2 mmol), N-hydroxysuccinimide (0.35 g, 3 mmol), and DCC (0.4 g, 2 mmol) were dissolved in 20 mL of dichloromethane. After the reaction was refluxed under N2 protection for 12 h, the resulting suspension was filtered out and dichloromethane was then evaporated under vacuum. The residue was chromatographed on silica gel using dichloromethane/methanol (9:1) to give a white solid and recrystallized by ethyl acetate/n-hexane (6/1) to give 0.7 g of 6 (yield 65%). Mp 85-87 °C, 1H NMR (300 MHz, CDCl3): δ 8.21 (d, 1H), 7.39 (d, 1H), 7.12 (d, 1H), 5.60 (s, 2H), 3.61 (m, 1H), 3.08 (m, 2H), 2.99 (t, 2H), 2.89 (t, 2H), 2.82 (t, 4H), 2.62 (t, 2H), 2.50 (m, 1H), 1.95 (m, 1H), 1.75 (m, 4H), 1.56 (m, 2H); 13C NMR (500 MHz, CDCl3): δ 24.34, 25.53, 25.53, 26.23, 28.62, 28.66, 34.03, 34.51, 38.50, 40.24, 56.24, 63.20, 121.71, 121.76, 126.96, 134.36, 144.12, 154.59, 167.57, 168.82, 168.82, 170.23, 170.92. Anal. Calcd: C 49.82, H 4.69, N 5.05, S 11.55, O 28.88, Found: C 50.05, H 4.69, N 5.17, S 11.08, O 28.82; MS (APCI): 589 (SSTN + Cl). Instruments and Methods. TLC was performed with Analtech silica gel FG TLC plates (250 µm). Flash chromatography was carried out using Aldrich silica gel (60 mesh). 1H NMR spectra were measured on a GE NMR-OMEGA (300 MHz), and 13C NMR spectra were

Bioconjugate Chem., Vol. 15, No. 5, 2004 1031

measured on a 500 MHz Bruker DMX. Elemental analysis was done by Atlantic Microlab Inc., Atlanta, GA. Mass spectra (GC/MS) were recorded at the University of Chicago. A UV-vis absorption spectrometer was used to follow the changes in absorbance of SSTN fragmentation products during the photolysis. The UV-vis spectra were measured on a Hewlett-Packard 8453A diode-array spectrophotometer with a wavelength range from 190 to 1100 nm and a 1 nm bandwidth. The steady-state emission spectral measurements were carried out using 1 cm × 1 cm quartz cells in a PTI C60 spectrofluorimeter with a 75 W mercury lamp as an irradiation source. The detector was a thermoelectrically cooled photomultiplier. A rightangle configuration was used for emission and excitation. The protein-immobilized substrate was placed at an angle of 30° with respect to the excitation light in order to reduce the reflection of excitation light. The surface was covered with 30 µL of PBS during the measurement. A multimode Nanoscope IIIa AFM (Digital Instruments, Santa Barbara, CA) was used to carry out and examine protein delivery. BSA or anti-human IgG antibody was immobilized on a gold-coated AFM tip via SSTN. The gold coating of the AFM cantilever was created using a thermal evaporator (Denton Vacuum, Moorestown, NJ). A gold-coated AFM tip (3 nm precoated titanium and 50 nm gold) was immersed in a 1.64 × 10-4 M SSTN solution in DMF at room temperature under nitrogen protection, generating a monolayer of SSTN on the tip via Au-S linkage, as routinely performed in the lab (15, 16). After being thoroughly rinsed, the modified tip was immersed in a 10-6 M BSA or anti-human IgG antibody solution (in PBS buffer) overnight at room temperature under nitrogen protection. This allowed the chemical bonding of proteins onto the AFM tip. The unbound proteins were removed by thoroughly rinsing the tip in PBS and deionized water. For local functionalization of proteins at the tip end, we used an SSTNmodified AFM tip to approach a thin layer of protein solution on a mica surface. The approaching process was manually controlled using the positioner of the AFM setup and was carefully monitored using the optical microscope equipped with the AFM to allow only the tip end to contact the protein solution. After a 1-h incubation, the tip was withdrawn and thoroughly rinsed in PBS and deionized water to remove any unbound residue. A fluorescence microscope (Olympus IX-70) was used to confirm the local protein functionalization when a fluorescein-labeled anti-human IgG was employed. Photocontrollable protein delivery was performed on either a freshly cleaved mica microscope cover slide surface or a thoroughly cleaned glass microscope cover slide surface. AFM images were acquired in a contact mode using a triangular Si3N4 tip. Photochemical Reactions. Photolysis was performed with a 3.6 × 10-4 M SSTN solution in CH3CN using a quartz cuvette. The irradiation was carried out at room temperature using a 100 W high-pressure mercury lamp (Oriel Model 68700) with a UV 297 interference filter (Andover Corporation Optical Filter). Irradiation timedependent photolysis was monitored by periodically taking the absorption spectra of SSTN to quantify the reaction quantum yield. To identify the photoreaction products, the sample was irradiated for 40 min. After the solvents were removed under vacuum, the residues were chromatographed on silica gel using chloroform to give yellow 7 in gel form (0.04 g, 60%). 1H NMR (300 MHz, CDCl3): δ 9.85 (s, 1H), 8.12 (d, 1H), 7.45 (d, 1H), 7.05 (d,

1032 Bioconjugate Chem., Vol. 15, No. 5, 2004

Yan et al.

Scheme 1. Synthetic Scheme for Succinic Acid Succinimidyl Ester 5-Thioyloxy-2-nitrobenzyl Ester (SSTN)a

a Solvents and catalysts: (a) THF, p-toluenesulfonic acid, 86%; (b) THF, DCC, 85%; (c) 0.2 M HCl/H O/THF (1:2 v/v) rt, 90%; (d) 2 CH2Cl2, DMAP, rt, 75%; (e) CH2Cl2, DCC, rt, 65%.

1H), 3.61 (m, 1H), 3.08 (m, 2H), 2.62 (t, 2H), 2.50 (m, 1H), 1.95 (m, 1H), 1.70 (m, 4H), 1.56 (m, 2H). Actinometry and Quantum Yield Measurement. Potassium ferrioxalate actinometry was used as a reference to determine the incident light intensity (17). Potassium ferrioxalate was recrystallized from warm water in the dark and dried in a vacuum oven. The concentration of the ferrioxalate was 0.006 M in water. The photoreaction of potassium ferrioxalate was followed by UV-visible absorption. The amount of the product can be directly determined and then converted to determine the quantity of photons delivered (in Einstein). The quantum yield of the photoreaction of SSTN in acetonitrile can then be calculated using the following equation and the lamp flux value:

φ)

moles of product × C lamp flux × irradiation time (s)

The amount of product is determined by the absorbance change from UV-vis. C is a value that corrects the incomplete light absorption at the excitation wavelength. C ) 1/(1 - T), where T is the transmittance of SSTN before the irradiation. Photocleavable Protein Immobilization. A microscope cover slide (Fisher) was used as a substrate. After being cleaned with H2O2/H2SO4 solution, the slide was coated with a 100 nm layer of gold on top of 10 nm of precoated titanium. The gold-coated slides were then immersed into SSTN/CH3CN solution (3.6 × 10-4 M) overnight at room temperature under nitrogen protection (5). After being thoroughly rinsed using CH3CN and then water, the substrates were then incubated in avidin solution (0.5 mg/mL in 0.1 M NaHCO3, pH 8.5) for 4 h at room temperature. After being rinsed with PBS buffer, the avidin-immobilized substrates were immersed in an Alexa 488 biotin solution (2.56 × 10-4 M) for 1 h, followed by multiple rinsings of the substrate to remove any unbound biotin. The immobilization of avidin on the gold substrate surface was confirmed by the strong fluorescence of the coverslips (data not shown). To investigate the photocleavage of avidin from the substrate, emission spectra of the protein-functionalized surface after irradiation and thorough washing were recorded at 2-min intervals in PBS buffer solution. A control experiment was carried out by replacing SSTN with the conventional

cross-linker succinimidyl 3-2(2-pyridyldithio)propionate (SPDP) under the same experimental conditions. RESULTS AND DISCUSSION

Synthesis of Succinic Acid Succinimidyl Ester 5-Thioyloxy-2-nitrobenzyl Ester. The synthesis of succinic acid succinimidyl ester 5-thioyloxy-2-nitrobenzyl ester (SSTN, 6) was achieved using 5-hydroxy-2-nitrobenzyl alcohol and thioctic acid (1,2-dithiolane-3-pentanoic acid) as starting materials (Scheme 1). The main challenge of this synthetic route was to selectively synthesize 4 without the competitive reactions of the two hydroxyl groups presented in 1 with thioctic acid. This was achieved using 3,4-dihydro-2H-pyran as the specific primary hydroxyl protection reagent. Since electrons on phenol oxygen are conjugated with π electrons and are strongly withdrawn by the nitro group, the electron density of the phenol group is much higher than that of the primary hydroxyl group. Therefore, the nucleophilic addition reaction in the first step should be highly selective toward the primary hydroxyl group when compared to the phenol group (18). In brief, the primary hydroxyl group of 1 was selectively protected by reacting with 3,4-dehydro-2H-pyran (19, 20) in the THF in the presence of p-toluenesulfonic as the catalyst (yield of 86%). Esterification of the phenol group with thioctic acid was carried out in THF using DCC as the catalyst and yielded nitrobenzyl thioctate derivative 3 (yield of 85%). Deprotection of the hydroxyl group was achieved in dilute hydrochloride H2O/THF solution (0.2 M, 1:2 V/V), and then tetrahydropyran was removed to yield the primary hydroxyl group 4, which further reacted with succinic anhydride in dichloromethane to yield 75% of hemisuccinic acid 5. The final product was synthesized by esterifying 5 with N-hydroxysuccinimide in dichloromethane, using DCC as a catalyst with a yield of 65%. The overall yield was 32%. The structures of the final compound and the intermediate products were confirmed by 1H NMR, 13C NMR, mass spectrometry, and elemental analysis. Photophysics and Photochemistry of the CrossLinker. Steady-state spectroscopic studies revealed that SSTN contains a major absorption peak (n-σ*) at 274 nm. As with other nitro-aromatic molecules, SSTN has a very weak fluorescence peak at 427 nm. It has been established that R-substituted 2-nitrobenzyl photophore

Photocleavable Cross-Linker

Bioconjugate Chem., Vol. 15, No. 5, 2004 1033

Scheme 2. Photolysis of SSTN

can undergo a highly efficient and fast photofragmentation reaction (21). Indeed, upon irradiation (λ ) 297 nm) of 7 mM of SSTN in deuterated chloroform, a thin-layer chromatograph experiment indicated that two major products were formed when visualized in an iodine chamber. To identify the photoreaction products, the reaction was followed using NMR spectroscopy. Benzyl hydrogen presented in SSTN exhibited a characteristic singlet peak at 5.60 ppm, which disappeared when the SSTN solution (3.6 × 10-4 M) was irradiated with 297 nm light. Another characteristic singlet peak centered at 9.85 ppm emerged as the photoreaction proceeded, suggesting a clean C-O bond cleavage and the formation of nitrosobenzaldehyde and succinic acid monosuccinimidyl ester. On the basis of TLC and NMR experimental results, we conclude that the reaction pathway is as shown in Scheme 2. The reaction scheme matches the reported reaction mechanism of the photolysis of other similar R-substituted 2-nitrobenzyl compounds (21). The photofragmentation reaction can also be monitored using UV-vis absorption spectroscopy. As shown in Figure 1, the product nitrosobenzaldehyde 7 has a π-π* transition centered at 334 nm (the other product 8 has no absorption peak above 300 nm), 60 nm red-shifted when compared to the major absorption peak at 274 nm from the starting material. The occurrence of the photofragmentation reaction was made evident by the decrease of SSTN absorption at 274 nm and the increase of nitrosobenzaldehyde absorption at 334 nm.

Figure 1. UV-vis spectra of 6 (3.0 × 10-4 M in acetonitrile) (a) and 7 (2.2 × 10-4 M in acetonitrile) (b).

Figure 2 demonstrates the absorption spectrum change of SSTN after the reaction solution (SSTN in acetonitrile, 9.0 × 10-5 M) was irradiated by 297 nm light for a short irradiation time (2-min interval). The insert of Figure 2 shows the absorption spectrum change of the SSTN solution (3.6 × 10-4 M) when irradiated for a longer time (30 min). The presence of an isosbestic point at 298 nm indicates a clean conversion from o-nitrobenzyl ester to nitrosobenzaldehyde. The progress of the reaction was followed quantitatively by noting the appearance of nitrosobenzaldehyde, as indicated by an absorption peak at 334 nm. A standard working curve for the absorption of pure nitrosobenzaldehyde as a function of concentration in the presence of the same concentration of starting material was used as a reference to quantify the product. The result of a control experiment, in which pure 7 in

acetonitrile was irradiated using the same light source, suggests that 7 is very stable under the photoreaction conditions. Thus, the increase at 334 nm is correlated to the formation of 7. After the photoflux using potassium ferrioxalate actinometry was determined, the quantum yield of the photoreaction with excitation at 297 nm was calculated to be 0.10 ( 0.04, comparable to the value reported in the literature (22). It was also found that the quantum efficiency remains the same in the degassed solvent; thus, oxygen most likely does not play any role in the photofragmentation reaction.

Figure 2. Absorption spectra of SSTN (9.0 × 10-5 M) under photolysis with shorter time interval. a: before irradiation; b: 2 min irradiation; c: 4 min irradiation; d: 6 min irradiation. Insert absorption spectra of SSTN (3.6 × 10-4 M) under photolysis with longer time interval. e: before irradiation; f: 18 min irradiation; g: 24 min irradiation; h: 30 min irradiation.

The mechanism of photolysis of o-nitrobenzyl alcohol derivatives (esters) involves a photoinduced intramolecular hydrogen abstraction and redox rearrangement that forms an o-quinonoid intermediate, followed by the release of the carboxylic acid and the formation of nitrosobenzaldehyde (23). Thus, the low quantum yield observed for the photoreaction is due to the low efficiency of the intramolecular hydrogen abstraction. Two approaches can be used to further increase photoefficiency: (1) introduce an additional o-nitro group to improve the chances of hydrogen abstraction by the excited nitro group; (2) introduce an R substitute to the photophore to enhance the efficiency of hydrogen abstraction and thus increase the overall reaction efficiency. Both approaches involve more steps to synthesize the desired cross-linkers. These approaches will be explored in our future study. Photocontrollable Protein Surface Modification. We used the biotin-avidin model system to demonstrate phototunable protein immobilization using the SSTN cross-linker. The extremely high affinity of the biotin molecule to avidin protein (association constant of 1015 M-1) has been well documented in the literature (24). As illustrated in Figure 3a, a monolayer of SSTN was selfassembled onto a gold-coated glass slide (25, 26). Nonspecifically bound SSTN was extensively washed away. After 4 h of incubation of the modified surface with avidin, the proteins were conjugated on the surface by the reaction of the amine groups of lysines in the avidin

1034 Bioconjugate Chem., Vol. 15, No. 5, 2004

Yan et al.

Figure 3. A. Scheme of photocontrollable protein immobilization. B. Fluorescence spectra of the substrates anchored with Alexa 488 biotin-avidin complex by photocleavable SSTN cross-linker, a: before irradiation; b: 2 min irradiation; c: 30 min irradiation.

with hydroxysuccinimidyl in the SSTN to form amide bonds. Again, any nonspecifically bound avidin was extensively washed away. As a control experiment, we also anchored avidin to a gold-coated substrate using succinimidyl 3-2(2-pyridyldithio)propionate (SPDP), a conventional cross-linker. Both substrates were then exposed to the Alexa-488-biotin solution (0.5 mg/mL) for 1 h, and unbound species were washed away with PBS buffer. Emission spectra of both substrates were recorded. The strong fluorescence at 520 nm indicated the formation of biotin-avidin complex on both substrates. Subsequently, both substrates were irradiated with UV light (297 nm) at 2-min intervals, followed by thorough washing steps (three times) before the recording of the fluorescence spectra of the substrates in the PBS buffer solution. Figure 3b demonstrates the changes to the normalized fluorescence intensities of the SSTN-immobilized substrate with different illumination times. The SSTN-immobilized surface lost its fluorescence almost completely after 30 min of UV irradiation. On the other hand, the fluorescence of the SPDP-immobilized surface decreased less than 10% after 30 min of UV irradiation and maintained this level with longer irradiation times (data not shown here). Consistent with the result of photolysis of SSTN (shown in Figure 2), we conclude that the decrease of the fluorescence on the substrate is due to the break of the C-O bond of SSTN and the release of avidin-biotin-dye complex from the substrate to the solution. This is confirmed by the maintained, strong fluorescence on the SPDP-immobilized substrate even after long-term irradiation. Thus, we can establish that SSTN is an efficient photocleavable cross-linker that can be used to attach and detach protein molecules on a surface in a controllable fashion. Importantly, the 297 nm UV light does not alter the binding affinity between biotin and avidin, as demonstrated in the control experiment. Photocontrollable Protein Delivery. Photocontrollable protein immobilization on an AFM tip provides an opportunity to deliver protein molecules to a desired location on a substrate. As shown in Figure 4a, we anchored BSA on a gold-coated silicon nitride cantilever using an SSTN cross-linker. The surface of the freshly cleaved mica was imaged using this tip in air contact mode. As shown in Figure 4b, the surface is flat and lacks any features. While the tip was maintained at the same location, UV irradiation (297 nm) was applied to the cantilever for 30 min. The same area of mica surface was then imaged by the same tip (image is shown in Figure 4c). Numerous particles with an average diameter of 50 ( 10 nm and average height of 4 ( 1 nm appeared on

Figure 4. (a) Schematic illustration of AFM tip functionalization with BSA and photocleavage of BSA from the tip. (b) 5 µm × 5 µm contact mode image of newly cleaved mica surface before UV irradiation. (c) 5 µm × 5 µm image collected at the same region with the same tip right after UV irradiation for 1 h. (d) 18 µm × 18 µm image under the same conditions as c.

the surface. A larger scale image (Figure 4d) illustrates the distribution of the particles with uniform size. Note that the wavy parallel lines are result of optical interference due to the flatness of the mica surface and are not surface-relevant features. A control experiment was performed under the same conditions, except that UV irradiation was removed by keeping the sample in the dark for the same time period. In this case, a featureless mica surface was revealed using the AFM. This indicates the particles we observed in Figure 4c are BSA delivered from the AFM cantilever upon photocleavage of SSTN. BSA has an ellipsoidal structure with a short axis of 4 nm and a long axis of 14 nm. Thus, the particles appearing on the mica are clusters of single-layered BSA lying down on the substrate. Each cluster contains approximately 35-50 BSA molecules. High-resolution images indicated that there were no individual protein molecules present on the surface other than the clusters. To further confirm that the species dropped on the substrate are proteins, we used fluorescein (FITC)labeled anti-human IgG antibody to repeat the experi-

Photocleavable Cross-Linker

Bioconjugate Chem., Vol. 15, No. 5, 2004 1035

Figure 7. Single BSA cluster delivered by AFM tip (4 µm × 4 µm).

Figure 5. Fluorescence image of clusters of fluorescein-labeled anti-human IgG delivered from an AFM tip.

Figure 6. (a) Bright field image of an AFM tip; (b) fluorescence image of the AFM tip modified with FITC-labeled anti-IgG.

ment. Again, the AFM images showed that the antibody fell to the substrate as clusters with a uniform diameter of 70 ( 10 nm. Since the approximate size of intact IgG is 15 nm × 8 nm, each cluster may contain ∼32-42 IgG molecules. Taking advantage of the fluorescein label of the antibody molecules, we were able to identify the protein clusters on the substrate using the fluorescence image, as shown in Figure 5. It was a surprise to observe that proteins fell to the substrate as uniform clusters. We propose that after photocleavage, the proteins remain on the surface of the cantilever due to surface adhesion. Protein-protein attraction drives the proteins to accumulate and form clusters. When the proteins accumulate to a certain size, the gravity of the protein cluster is comparable to the adhesive force and the cluster falls to the substrate. A detailed mechanism is currently under investigation in our labs. Because tip modification was achieved by incubating the cantilever in protein aqueous solution, proteins were immobilized on the entire cantilever (Figure 4a). When UV irradiation was applied, numerous protein clusters were delivered onto the surface. To accurately control the location of protein delivery, we modified protein molecules exclusively at the tip end. This was achieved by precisely controlling the tip to approach the protein solution, allowing only the tip end to contact the solution for incubation. The success of local tip modification was confirmed by examining the fluorescence image of a tip modified by fluorescein-labeled anti-human IgG antibody in this manner. Figure 6 shows the optical image (Figure 6a) and the fluorescence image (Figure 6b) of the AFM tip on its cantilever, achieved subsequently at the fixed tip position. Fluorescence was detected exclusively at the

tip end. After this tip was exposed to UV irradiation, a single protein cluster was observed at the position right underneath the tip, as shown in the AFM image (Figure 7) acquired using the same tip right after protein photocleavage. The conclusion is 2-fold: (1) proteins can be precisely delivered to a desired position; (2) an AFM tip can serve for both protein delivery and local imaging to examine the consequence of protein delivery. A challenging task in bioscience is to directly monitor a protein function at local regions in their natural environment. This creates an urgent requirement for a general approach for delivering a ligand to a precise position, with a precise quantity. Our protein delivery approach employed UV light as a remote switch to control both timing and the position of protein delivery. To the best of our knowledge, this is the only approach that can offer the capability to deliver uniform small granules of proteins to a substrate. We also demonstrated that the same tip can be used after photocleavage to image the same local area, which is critical to achieving information on the nanometer scale. Using this approach, we expect to be able to deliver ligand clusters to a cell surface. Guided by the specific interaction via force mapping (15, 16, 27) using the ligand-functionalized tip, cell membrane receptors will be registered. The ligand will then be delivered to a receptor-rich area in order to in-situ monitor the living cell’s localized response to the ligandreceptor binding on the nanometer scale. The feasibility has been demonstrated by our preliminary study in a model system involving the interaction between cholera toxin B oligomer (CTB) and its receptor ganglioside GM1. This approach will open an avenue for answering questions such as how individual ligand-receptor interactions trigger the cell signaling pathways in a living cell and why an inhibitor needs to be delivered to certain protein complexes in order to block protein activity. CONCLUSION

A photocleavable cross-linker, succinic acid succinimidyl ester 5-thioyloxy-2-nitrobenzyl ester (SSTN), was synthesized in five steps using 5-hydroxy-2-nitrobenzylalcohol and thioctic acid (1,2-dithiolane-3-pentanoic acid) as starting materials. Upon photoexcitation, SSTN undergoes a fragmentation reaction to produce carboxylic acid and nitrosobenzaldehyde with a quantum yield of 0.1. SSTN can covalently anchor protein molecules to a gold-coated substrate, and those protein molecules can subsequently be efficiently (>95%) released under UV irradiation. By immobilizing the protein molecules on a gold-coated AFM tip, the protein molecules can be delivered as uniform clusters to the substrate via photolysis, and subsequently imaged at the same local region. Thus, this approach provides an important tool for tunable surface modification and precise protein delivery on the nanometer scale. We also expect that the results obtained in this study will lead to a new avenue

1036 Bioconjugate Chem., Vol. 15, No. 5, 2004

for generating protein patterns. In this case, the crosslinker will function as a “photoresister”. When irradiation is applied to the protein-functionalized surface through a conventional photomask, as used in photolithography, the corresponding protein patterns can be easily generated. ACKNOWLEDGMENT

F. Yan thanks the support from Kilpatrick fellowship of IIT. We also thank Dr. Brian Kay at Argonne National Laboratory for reviewing the manuscript and insightful comments. This work is supported by ANL LDRD and NSF (IBN-0103080). LITERATURE CITED (1) For review, Subrahmanyam, S., Piletsky, S. A., and Turner, A. P. F. (2002) Application of natural receptors in sensors and assays. Anal. Chem. 74 (16), 3942-3951. (2) Rao, S. V., Anderson, K. W., and Bachas, L. G. (1998) Oriented immobilization of proteins. Mikrochim. Acta 128, 127-143. (3) Kindermann, M., George, N., Johnsson, N., and Johnsson, K. (2003) Covalent and Selective Immobilization of Fusion Proteins. J. Am. Chem. Soc. 125 (26), 7810-7811. (4) Cao, R., Gu, Z., Patterson, G. D., and Armitage, B. A. (2004) A recoverable enzymatic microgel based on biomolecular recognition. J. Am. Chem. Soc. 126 (3), 726-727. (5) For review, Dorma´n, G., and Prestwich, G. D. (2002) Using photolabile ligands in drug discovery and development. Trends Biotechnol. 18, 64-77. (6) Smet, M., Liao, L. X., Dehaen, W., and Dominic, V. M. (2000) Photolabile dendrimers using o-nitrobenzyl ether linkages. Org. Lett. 2 (4), 511-513. (7) Sternson, S. M., and Schreiber, S. L. (1998) An acid- and base-stable o-nitrobenzyl photolabile linker for solid-phase organic synthesis. Tetrahedron Lett. 39, 7451-7454. (8) Holmes, C. P., and Jones, D. G. (1995) Reagents for combinatorial organic synthesis: development of a new o-nitrobenzyl photolabile linker for solid phase synthesis. J. Org. Chem. 60, 2318-2319. (9) Teague, S. (1996) Facile synthesis of a o-nitrobenzyl photolabile linker for combinatorial chemistry. Tetrahedron Lett. 37 (32), 5751-5754. (10) Rock, R. S., and Chan, S. I. (1996) Synthesis and photolysis properties of a photolabile linker based on 3′-methoxybenzoin. J. Org. Chem. 61, 1526-1529. (11) Olejnik, J., Sonar, S., Krzyman˜sk-Olejnik, E., and Rothschild, K. J. (1995) Photocleavable biotin derivatives: A versatile approach for the isolation of biomolecules. Proc. Natl. Acad. Sci. U.S.A. 92, 7590-7594. (12) Ottl, J., Gabriel, D., and Marriott, G. (1998) Preparation and photoactivation of caged fluorophores and caged proteins using a new class of heterobifunctional, photocleavable crosslinking reagents. Bioconjugate Chem. 9 (2), 143-151.

Yan et al. (13) Rozhok, S., Piner, R., and Mirkin, C. A. (2003) Dip-pen nanolithography: what controls ink transport? J. Phys. Chem. B. 107 (3), 751-757. (14) Pena, D. J., Raphael, M. P., and Byers, J. M. (2003) “Dippen” nanolithography in registry with photolithography for biosensor development. Langmuir 19 (21), 9028-9032. (15) Wang, R., Vengasandra, S., and Yan, F. (2001) Probing the nature of cholera toxin b oligomer by the atomic force microscopy. Chem. Lett. 11, 1170-1171. (16) Vengasandra, S., Sethumadhavan, G., Yan, F., and Wang, R. (2003) Studies on the protein-receptor interaction by atomic force microscopy. Langmuir 19 (26), 10940-10946. (17) Murov, S. L., Carmichael, I., and Hug, G. L. (1993) Handbook of Photochemistry, 2nd ed.,pp 298-305, Marcel Dekker Inc., New York. (18) Smith, M. B., and March, J. (2001) March’s Advanced Organic Chemistry, 5th ed.,pp 994-998, Wiley-Interscience, New York. (19) Yu, J., Orfino, F. P., and Holdcroft, S. (2001) Structural order in conjugated organic films prepared by catalytic deprotection of self-assembled polymers. Chem. Mater. 13, 526-529. (20) Yu, J., and Holdcroft, J. (2000) Solid-state thermolytic and catalytic reactions in functionalized regioregular polythiophenes. Macromolecules 33, 5073-5079. (21) Pearson, A. J., and Roush, W. R. (1999) Handbook of Reagents for Organic Synthesis: Activating Agents and Protecting Groups, pp 287-289, John Wiley & Sons Ltd., New York. (22) Houlihan, F. M., Shugard, A., Gooden, R., and Reichmanis, E. (1988) Nitrobenzyl ester chemistry for polymer processes involving chemical amplification. Marcromolecules 21, 20012006. (23) Yip, R. W., Sharma, D. K., Giasson, R., and Gravel, D. (1985) Photochemistry of the o-nitrobenzyl system in solution: evidence for singlet-state intramolecular hydrogen abstraction. J. Phy. Chem. 89, 5328-5330. (24) Laitinen, O. H., Marttila, A. T., Airenne, K. J., Kulik, T., Livnah, O., Bayer, E. A., Wilchek, M., and Kulomaa, M. S. (2001) Biotin Induces tetramerization of a recombinant monomeric avidin. J. Biol. Chem. 276, 8219-8224. (25) Maya, L. (2000) Polymer-mediated assembly of gold nanoclusters. Langmuir 16, 9151-9154. (26) Dong, Y., and Shannon, C. (2000) Heterogeneous immunosensing using antigen and antibody monolayers on gold surfaces with electrochemical and scanning probe detection. Anal. Chem. 72, 2371-2376. (27) Gad, M., Itoh, A., and Ikai, A. (1997) Mapping cell wall polysaccharides of living microbial cells using atomic force microscopy. Cell Biol. Int. 21, 697-706. (28) Chada, V. G. R., Malinowska, K., Menhart, N., and Wang, R. Identification of TrkA on living PC12 cells by the atomic force microscopy, Biochim. Biophys. Acta, submitted.

BC049901D