Anal. Chem. 2005, 77, 5089-5095
Catalytic Three-Dimensional Protein Architectures Richard Allen,† Rex Nielson,† Dana D. Wise, and Jason B. Shear*
Department of Chemistry & Biochemistry and The Institute for Cellular & Molecular Biology, 1 University Station A5300, The University of Texas, Austin, Texas 78712
We demonstrate a strategy for microfabricating catalytically active, three-dimensional matrixes composed of cross-linked protein in cellular and microfluidic environments. In this approach, a pulsed femtosecond laser is used to excite photosensitizers via multiphoton absorption within three-dimensionally defined volumes, a process that promotes cross-linking of protein residue side chains in the vicinity of the laser focal point. In this manner, it is possible to fabricate protein microparticles with dimensions on the order of the multiphoton focal volume (less than 1 µm3) or, by scanning the position of a laser focal point relative to a specimen, to generate surface-adherent matrixes or cables that extend through solution for hundreds of micrometers. We show that protein matrixes can be functionalized either through direct cross-linking of enzymes, by decoration of avidin matrixes with biotinylated enzymes, or by cross-linking biotinylated proteins that then are linked to biotinylated enzymes via an avidin couple. Several formats are explored, including microparticles that can be translocated to desired sites of action (including cytosolic positions), protein pads that generate product gradients within cell cultures, and oncolumn nanoreactors for microfluidic systems. These biomaterial fabrication technologies offer opportunities for studying a variety of cell functions, ranging from singlecell biochemistry and development to perturbation and analysis of small populations of cultured cells.
Cells exist in complex three-dimensional (3D) microenvironments where subtle chemical changes direct processes such as differentiation and polarized cell growth. Diffusible and immobilized gradients of bioactive molecules released from neighboring cells or cells at distant locations form developmental cues vital to tissue function and to the fundamental architecture of an organism.1,2 The response of a cell to these evolving biochemical surroundings is often governed by heterogeneous intracellular domains that exist only transiently in nonmembrane-bound cytosolic regions.3-5 * To whom correspondence should be addressed. E-mail: jshear@ mail.utexas.edu. † These authors contributed equally to this work. (1) Tessier-Lavigne, M.; Goodman, C. S. Science 1996, 274, 1123-1133. (2) Song, H. J.; Poo, M. M. Curr. Opin. Neurobiol. 1999, 9, 355-363. (3) Wong, W.; Scott, J. D. Nat. Rev. Mol. Cell. Biol. 2004, 5, 959-970. (4) Hartwell, L. H.; Hopfield, J. J.; Leibler, S.; Murray, A. W. Nature 1999, 402, C47-C52. (5) Pines, J. Nat. Cell. Biol. 1999, 1, E73-E79. 10.1021/ac0507892 CCC: $30.25 Published on Web 07/15/2005
© 2005 American Chemical Society
Various strategies that produce spatiotemporal chemical gradients have been developed for investigating these processes in model culture systems. In some instances, two-dimensional guidance can be achieved through precision surface patterning of biomolecules, an approach that has been used to study growth cone pathfinding.6-8 Modification of extracellular environments in three dimensions also is possible through directed solution delivery (e.g., via micropipets9 and microfluidics10,11) and highresolution manipulation of chemically defined microspheres.12 Although useful for some applications, these approaches often provide insufficient precision or flexibility to affect cellular chemistry at multiple positions with the resolution necessary to investigate spatially polarized responses. Introduction of membraneimpermeable chemicals to the cytosol, generally accomplished using electroporation,13,14 photorelease,15 and microinjection,16 can localize effectors to desired subcellular regions on short time scales, but typically are limited over longer periods by diffusion. To address such limitations, we have developed a technique based on multiphoton-excited (MPE) protein cross-linking to fabricate chemically active structures with micrometer 3D resolution in cellular environments. Until recently, microarchitectures fabricated using MPE photochemistry were based on synthetic resins and were fashioned into structures such as exquisitely detailed (but chemically inert) artwork,17,18 high-aspect-ratio towers,19 and micromachines.20 As a biomolecular alternative, Cam(6) Rosoff, W. J.; Urbach, J. S.; Esrick, M. A.; McAllister, R. G.; Richards, L. J.; Goodhill, G. J. Nat. Neurosci. 2004, 7, 678-682. (7) Kane, R. S.; Takayama, S.; Ostuni, E.; Ingber, D. E.; Whitesides, G. M. Biomaterials 1999, 20, 2363-2376. (8) Whitesides, G. M.; Ostuni, E.; Takayama, S.; Jiang, X.; Ingber, D. E. Annu. Rev. Biomed. Eng. 2001, 3, 335-373. (9) Lohof, A. M.; Quillan, M.; Dan, Y.; Poo, M. M. J. Neurosci. 1992, 12, 12531261. (10) Takayama, S.; Ostuni, E.; LeDuc, P.; Naruse, K.; Ingber, D. E.; Whitesides, G. M. Chem. Biol. 2003, 10, 123-130. (11) Takayama, S.; Ostuni, E.; LeDuc, P.; Naruse, K.; Ingber, D. E.; Whitesides, G. M. Nature 2001, 411, 1016. (12) Zhang, X.; Poo, M. M. Neuron 2002, 36, 675-688. (13) Haas, K.; Sin, W. C.; Javaherian, A.; Li, Z.; Cline, H. T. Neuron 2001, 29, 583-591. (14) Olofsson, J.; Nolkrantz, K.; Ryttsen, F.; Lambie, B. A.; Weber, S. G.; Orwar, O. Curr. Opin. Biotechnol. 2003, 14, 29-34. (15) Pettit, D. L.; Wang, S. S.; Gee, K. R.; Augustine, G. J. Neuron 1997, 19, 465-471. (16) Korzh, V.; Strahle, U. Differentiation 2002, 70, 221-226. (17) Serbin, J.; Egbert, A.; Ostendorf, A.; Chichkov, B. N.; Houbertz, R.; Domann, G.; Schulz, J.; Cronauer, C.; Frohlich, L.; Popall, M. Opt. Lett. 2003, 28, 301-303. (18) Kawata, S.; Sun, H. B.; Tanaka, T.; Takada, K. Nature 2001, 412, 697698. (19) Baldacchini, T.; LaFratta, C.; Farrer, R. A.; Teich, M. C.; Saleh, B. E. A.; Naughton, M. J.; Fourkas, J. T. J. Appl. Phys. 2004, 95, 6072-6076.
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005 5089
pagnola and co-workers demonstrated that similar structures could be created using two-photon excitation to cross-link proteins such as alkaline phosphatase and fibrinogen.21,22 Although these structures were shown to retain some native protein activity, the fabrication solutions contained toxic photosensitizers (e.g., rose bengal), limiting possibilities for application to living cellular environments.21 Rose bengal and methylene blue promote crosslinking mainly via singlet oxygen mechanisms (type II photosensitization) and can be both phototoxic23,24 and cytotoxic (in the absence of irradiation) at low micromolar levels.25 Naturally occurring flavins have been shown to function as photosensitizers, acting through both type II and oxygen-independent (type I) radical mechanisms to promote cross-links between photooxidizable residues (e.g., Tyr, Trp, Cys).26,27 Although such biological molecules can damage cells following photoexcitation, they largely avoid issues related to nonphotonic cytotoxicity. We recently have explored the use of flavins and other noncytotoxic biological compounds for sensitizing protein photo-cross-linking within cultures of living cells and have demonstrated the utility of this approach for influencing neuronal development in real time.28 Here, we investigate capabilities of this microfabrication strategy for localizing enzymatic reactions within cellular millieus and microanalytical environments, establishing the feasibility for creating chemical gradients within cell cultures and for probing low concentrations of biological molecules. Localization is achieved both through the direct photo-cross-linking of a desired enzyme and via more general strategies involving avidin-biotin conjugation chemistry. In addition to surface-adherent lines and platforms, the geometry of protein matrixes created using this approach ranges from micrometer-thick cables (tethered at their termini) to microparticles that are created within an optical trap and transported to a desired site of action. The wide range of chemical functionalies possible with this method opens important opportunities for characterizing and controlling intracellular and extracellular environments. EXPERIMENTAL SECTION Matrix Fabrication. The system used to generate photo-crosslinked protein structures in the presence of cultured cells has been described previously.27 Briefly, structures were produced via multiphoton-excited photochemistry using the output from a femtosecond titanium/sapphire (Ti:S) laser (Coherent Mira, Santa Clara, CA) operating at ∼750 nm. The Ti:S beam was aligned into a Zeiss Axiovert (inverted) microscope equipped with difference interference contrast (DIC) imaging. The beam was adjusted to approximately fill the back aperture of a 40× objective (Olympus (20) Galajada, P.; Ormos, P. Appl. Phys. Lett. 2001, 78, 249-251. (21) Pitts, J. D.; Howell, A. R.; Taboada, R.; Banerjee, I.; Wang, J.; Goodman, S. L.; Campagnola, P. J. Photochem. Photobiol. 2002, 76, 135-144. (22) Basu, S.; Wolgemuth, C. W.; Campagnola, P. J. Biomacromolecules 2004, 5, 2347-2357. (23) Kochevar, I. E.; Lynch, M. C.; Zhuang, S.; Lambert, C. R. Photochem. Photobiol. 2000, 72, 548-553. (24) Moan, J.; Peng, Q. Anticancer Res. 2003, 23, 3591-3600. (25) Paulino, T. P.; Magalhaes, P. P.; Thedei, G.; Tedesco, A. C.; Ciancaglini, P. Biochem. Mol. Biol. Ed. 2005 33, 46-49. (26) Spikes, J. D.; Shen, H. R.; Kopeckova, P.; Kopecek, J. Photochem. Photobiol. 1999, 70, 130-137. (27) Shen, H. R.; Spikes, J. D.; Smith, C. J.; Kopecek, J. J. Photochem. Photobiol., A 2000, 133, 115-122. (28) Kaehr, B.; Allen, R.; Javier, D. J.; Currie, J.; Shear, J. B. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 16104-16108.
5090
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
UPlanFl, 0.75 NA) with average powers of 50-100 mW, producing a nearly diffraction-limited focal volume. By scanning the position of the specimen relative to the laser focus using a motorized stage, extended structures of different dimensionality could be fabricated. For experiments coupling photo-cross-linking and optical tweezers, the Ti:S laser was operated in continuous wave mode at powers ranging from 1 to 1.5 W and aligned to overfill the back aperture of a 100× objective (Zeiss Fluar, 1.3 NA, oil immersion). Except where otherwise noted, matrixes were fabricated using solutions that contained protein (100-300 mg/mL) and flavin adenine dinucleotide (FAD) at 5 mM as a photosensitizer in Hepesbuffered saline (HBS, 10 mM Hepes, 135 mM NaCl, 5 mM KCl, 2 mM MgCl2, 2 mM CaCl2, 10 mM D-glucose, pH 7.4). Cables were fabricated using similar protein concentrations, but typically used 1 mM methylene blue (MB) as a photosensitizer due to its superior cross-linking abilities. (Although the specific photophysical and photochemical causes have not been delineated, we find that lines fabricated using MB are consistently thicker at a constant protein concentration and scan speed than are lines produced using FAD.) Chemicals and Materials. Except where noted, all chemicals were purchased from Sigma-Aldrich (St. Louis, MO). Bovine serum albumin (BSA) and biotinylated calf intestinal alkaline phosphatase (biotin-AP) were obtained from Equitech-Bio (Kerrville, TX) and Pierce Biotechnology, Inc. (Rockford, IL), respectively. Alexa 488-BSA, 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide (DiIC18(5)), and avidin were purchased from Molecular Probes, Inc. (Eugene, OR). Biotin-BSA Cables. BSA was biotinylated via reaction with excess 6-((biotinoyl)amino)hexanoic acid, succinimidyl ester (biotin-X, SE) in 100 mM carbonate buffer (pH 9.9) followed by dialysis against HBS to remove unreacted substrates. After fabrication, cables were functionalized with alkaline phosphatase (AP) by treatment with avidin (1.0 mg/mL; 2 min) followed by biotin-AP (0.1 mg/mL; 2 min) in HBS. Cell Culture. Neuroblastoma-glioma (NG108-15) cells were purchased from the American Type Culture Collection (Manassas, VA) and cultured in DMEM (Mediatech, Inc., Herndon, VA) supplemented with 10% fetal bovine serum (FBS), penicillin (100 mg/L), and streptomycin (100 kunits/L) purchased from Invitrogen (Carlsbad, CA). Flasks were maintained at 37 °C in a 10% CO2 atmosphere with saturated H2O. For all experiments, cells were seeded on 0.01% (w/v) poly(L-lysine)-coated glass coverslips and incubated for 1-3 days in a low-serum (1% FBS) growth medium. PDMS Microchamber Fabrication. To minimize the surface area subjected to nonspecific protein adsorption in extracellulargradient experiments, NG108-15 cultures were grown on coverslips that could be transferred to flow cells after fabrication of protein structures. In this approach, adapted from Folch et al.,29 a stencil with a 1.5-mm pore was formed from PDMS (10:1 RTV615A to RTV615B; GE Silicones, Niskayuna, NY) and was placed on a 22 × 22 mm coverslip (no. 1; Erie Scientific, Portsmouth, NH). This procedure created a well (∼1 mm deep) into which NG108-15 cells were seeded at ∼15 000 cells/mL. (29) Folch, A.; Jo, B. H.; Hurtado, O.; Beebe, D. J.; Toner, M. J. Biomed. Mater. Res. 2000, 52, 346-353.
For protein structure fabrication, medium was removed and a blocking solution of BSA (200 mg/mL) in HBS was added and incubated for 10 min before addition of photo-cross-linking solution (avidin and FAD). After photofabrication, the pore was washed extensively with HBS. To functionalize the avidin structure, the pore was filled with a 0.1 mg/mL solution of biotin-AP in HBS for 2 min and then rinsed several times with HBS. After the stencil was removed, a small droplet of HBS was left on the NG108-15 field, ensuring the cells remained submersed in buffer. An opensided channel fashioned from PDMS (see below) was aligned over the droplet, and the entire system was gently compressed in an aluminum clamping block to strengthen the seal between the coverslip and PDMS, after which the channel could be flooded with HBS and used as a flow cell. The negative master for the PDMS channel was prepared by affixing a section of a no. 1 1/2 coverslip (cut to the same dimensions as the desired channel) to the floor of a 3.5-cmdiameter Petri dish using epoxy. Teflon tubing (0.35-mm i.d., 0.65mm o.d.; Zeus TFE) was affixed to both ends of the channel master coverslip to create voids in PDMS through which solutions could be introduced into the channels containing cells. The Petri dish was filled with PDMS solution, degassed, and cured at 60 °C for 1 h. This process resulted in channels ∼170 µm high, 3.5 mm wide, and 20 mm long, which could be placed over cultures of NG108-15 cells. Cell Imaging. After matrix fabrication, initial inspection of structures, and placement of the flow cell, extracellular product gradients were generated and imaged on a Nikon Eclipse (TE2000E) microscope using a DAPI filter set. Confocal images of intracellular particles and protein cables were obtained using Leica (Wetzlar, Germany) confocal microscopes (models 4D and SP2 AOBS). Multiphoton NADH Measurements. The instrument used to obtain multiphoton measurements of enzymatic NADH production has been described in previous reports.30 Briefly, the output from a Ti:S (Verdi pumped Mira, Coherent, Inc.) operating at 735 nm was attenuated to ∼250 mW and focused to a nearly diffractionlimited spot using a 100× objective (Zeiss Fluar, 1.3 NA, oil immersion) into a chambered coverglass (Nalge Nunc International, Rochester, NY). Fluorescence generated in a threedimensionally resolved volume centered at the focal point was collected by the objective and was reflected from the laser axis using a visible dichroic mirror (Chroma Technology, Rockingham, VT; part 500DCRB). Residual scatter and other background was filtered using a cuvette (4-cm path length) filled with a saturated solution of CuSO4 and multiple BG-39 colored-glass filters (Andover Corp., Salem, NH), and signal was measured using a photomultiplier tube (HC125-02, Hamamatsu, Hamamatsu City, Japan) operated in photon-counting mode. NADH was produced and detected in the tip of square, 50-µm-i.d. fused-silica capillary (Polymicro Technologies, Phoenix, AZ). Capillary tip nanoreactors were prepared by focusing Ti:S light through the capillary sidewall (i.e., with the capillary axis perpendicular to laser propagation), enabling matrixes of glutamate dehydrogenase (GDH, type III, bovine liver) to be photo-crosslinked onto the surface of the channel wall. Typically, a structure (30) Gostkowski, M. L.; McDoniel, J. B.; Wei, J.; Curey, T. E.; Shear, J. B. J. Am. Chem. Soc. 1998, 120, 18-22.
was fabricated within a few centimeters of the capillary opening, which was cut to the appropriate location. A solution containing BSA (200 mg/mL), GDH (10 µg/mL), and FAD (5 mM) was used to fabricate glutamate nanoreactors. After extended washing (typically overnight) with 100 mM borate buffer, capillaries were filled with 100 mM phosphate buffer and positioned on the instrument. The capillary end containing a cross-linked GDH structure was positioned above the objective in solution within the chambered cover glass (aligned coaxially with the laser propagation axis), with the other capillary end submersed in a separate buffer reservoir. Care was taken to maintain equal meniscus heights in each reservoir during measurements to avoid hydrodynamic flow that would flush products from the measurement region. Above the objective, the capillary was aligned so that the laser focus was centered in the channel near the plane of the capillary aperture (i.e., within tens of micrometers of the GDH matrix). Reaction substrates (NAD+, 100 µM; glutamate, 100-300 nM; alanine, 100 µM; glutamic pyruvic transaminase (GPT), 4 µg/mL, ∼40 nM) were placed in the chambered cover glass and prevented from entering the capillary before the desired time point by raising the opposing buffer reservoir. Thirty second gravity injections (4-cm height differential) of reaction substrates (i.e., introduced from the chambered cover glass reservoir into the channel) were performed by manually lowering the opposing reservoir. Estimate of Protein Concentrations in Structures. BSA was coupled to the fluorescent probe, BodipyTR, through reaction with the STP ester (Molecular Probes) followed by dialysis to remove unreacted dye. The concentration of solution-phase BSA and its degree of labeling were determined through absorbance measurements at 590 and 280 nm using procedures described in the Molecular Probes Technical Bulletin, “Amine-Reactive Probes”. Protein matrixes were fabricated from a solution of BodipyTRBSA (100 mg/mL) and FAD (5 mM) on BSA-passivated glass coverslips. Data obtained from wide-field fluorescence images of these matrixes were compared to fluorescence microscopy measurements of solution-phase BodipyTR-BSA (known concentrations in a 50-µm-i.d. microchamber); by determining the thickness of protein matrixes using atomic force microscopy, estimates of matrix concentration were obtained. Sample Preparation for Electron Microscopy. Samples were prepared for electron microscopy by exchanging fabrication solutions with Tris-buffered saline (TBS; pH 8.5) containing 15% (w/v) glutaraldehyde. After incubation for 10 min, solution exchanges were performed using, in sequence, 100% H2O, 50% H2O/50% ethanol, 100% ethanol, 100% methanol (×2); each solution was allowed to incubate for 20 min before being replaced. The specimen was allowed to air-dry for 5 days after the final methanol wash, and the dry sample was coated using a Au/Pt target to a nominal thickness of 8 nm. RESULTS AND DISCUSSION Large-Aspect-Ratio Cross-Linked Structures. Cultured cells display differential patterns of development according to the physiochemical structure of their environment; experiments with 2D and 3D culture systems have demonstrated effects of cellcell and cell-matrix interactions on neuronal stem cell differentiation.31,32 The ability to alter the 3D cellular microenvironment through on-demand fabrication would extend the reach of tradiAnalytical Chemistry, Vol. 77, No. 16, August 15, 2005
5091
Figure 1. Large-aspect-ratio protein cables. (A) SEM of a BSA cable fabricated across a gap between coplanar borosilicate coverslips. SEM parameters: magnification, 453×; accelerating voltage, 4.0 kV; nominal working distance, 25 mm. Scale bar, 50 µm. Inset: Low magnification SEM image of the same specimen demonstrating a series of parallel cables. Although not apparent in this image, the second cable from the bottom sustained a small break during the SEM preparation procedure. (B) A series of diagonal Alexa-488labeled BSA cables fabricated between parallel coverslips spaced by ∼140 µm. Image represents reconstruction of a stack of ∼250 confocal fluorescence images spaced by ∼0.3 µm. (C) Catalytically active protein cables. A cable was fabricated from biotinylated BSA across an ∼180-µm gap between glass coverslips and was labeled with biotin-AP via an avidin sandwich approach. The sample then was treated with the fluorogenic AP substrate, 4-MUP (100-200 µM), which is catalytically converted to fluorescent 4-MU by AP. The intensity plot on the left was produced from a wide-field fluorescence image acquired under high flow conditions, in which steady-state product concentrations remain low. By reducing flow rates (right intensity plot), significant accumulation of 4-MU product can be attained. Background signal from the coverslips is caused by biotinAP retained as a result of nonspecific adsorption of biotin-BSA. Scale bar, 100 µm.
tional tissue engineering methods as well as our understanding of cells in biomimetic systems. To explore the capabilities of MPE cross-linking for fabricating large-aspect-ratio protein matrixes in various configurations, BSA cables were constructed that bridged gaps between coplanar or parallel glass surfaces spaced by distances greater than 100 µm (Figure 1). The thickness (cross-sectional diameter) of such cables is largely determined by laser power and the speed at which the laser focus is translated through the protein solution, with higher translation speeds and lower powers yielding thinner structures. At a scan speed of ∼10 µm/s and 100 mW average laser power, cables fabricated from BSA and MB displayed an average thickness of ∼2 µm. We also have fabricated cables using the photosensitizer FAD, although such structures generally require significantly greater care to avoid damage during washes. Figure 1C demonstrates the enzymatic activity of a functionalized protein cable under a flow of 4-methylum belliferyl phos(31) Liu, H.; Roy, K. Tissue Eng. 2005, 11, 319-330. (32) Silva, G. A.; Czeisler, C.; Niece, K. L.; Beniash, E.; Harrington, D. A.; Kessler, J. A.; Stupp, S. I. Science 2004, 303, 1352-1355.
5092 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
Figure 2. Product concentrations (4-MU) measured 20 µm downstream of an AP-functionalized cable as a function of solution flow rate. Point measurements were made using multiphoton excitation (Ti:S at 740 nm, 50 mW; 40×, 0.75 NA Olympus UPlanFl objective). The cable in this study (∼240 µm long suspended ∼180 µm from the floor of the flow cell) was fabricated from biotinylated BSA using multiple laser passes to improve durability. A volume flow rate of 1.00 mL/min in the flow cell used here corresponds to a linear flow rate of ∼30 mm/s at the cable. Error bars represent 1σ based on uncertainties in the linear regression of the calibration plot.
phate (4-MUP). This bridge, fabricated using a biotin-BSA conjugate, was decorated with biotin-AP via bridging avidin molecules. 4-MUP was catalytically converted to fluorescent 4-methylumbelliferone (4-MU) by biotin-AP immobilized to the cable. Accumulation of fluorescent product can be seen both in the vicinity of the cable and from coverslip-associated enzyme (the result of nonspecific biotin-BSA adsorption); as expected, product accumulation is inversely related to flow rates, an effect quantitatively evaluted on a separate cable structure (Figure 2). To evaluate whether protein cables were merely accumulating fluorescent product generated at the coverslip surface, control cables were fabricated using nonbiotinylated BSA alongside functionalized cables (data not shown); unlabeled structures displayed no significant 4-MU fluorescence. Mobile Microparticles. The possibility of using a photo-crosslinking laser to create and manipulate portable protein microparticlessquasi-one-dimensional structuressalso was explored. Here, tightly focused 800-nm CW light (∼800 mW at the back aperture of the objective) was used to excite multiphoton protein photo-cross-linking at a desired position in solution above the coverslip. Over a period of 60 s, a microparticle (diameter in the focal plane, ∼1 µm) could be developed at the focus and remain captured through optical trapping forces, providing a means for further manipulation. Microparticles fabricated in this manner conform approximately to a size and shape determined by the (nearly diffraction-limited) multiphoton voxel and the diffusion volume of reactive intermediates.33 After an initial formation period, the presence of a microparticle within the focal volume appears to inhibit further particle growth. Although it is feasible to exchange the fabrication solution for protein-free buffers while (33) Kawata, S.; Sun, H.-B. Mater. Res. Soc. Symp. Proc. 2003, 758, 163-168.
Figure 3. Portable protein microparticles. (A) A BSA microparticle is formed using tightly focused CW light from a titanium:sapphire laser (arrow point, left image), which also serves to retain the nascent particles via optical trapping. After fabrication, the microparticle is translocated through the plasma membrane of an NG108-15 cell. Scale bar, 5 µm. (B) A confocal section of another NG108-15 cell that has been loaded with a fluorescently labeled (Alexa 488) BSA microparticle (left edge of cell) in the manner described in part A. To evaluate the extent to which the particle was enclosed by the plasma membrane, the cell was labeled with the membrane-specific dye, DiIC18(5). Scale bar, 5 µm. (C) A intensity plot of a catalytically active microparticle composed of cross-linked avidin labeled with biotin-AP. Fluorescence imaging was performed under wide-field conditions. After formation in an optical trap, the avidin microparticle was transported to the wall of a fused-silica capillary, exposed to biotinAP (0.1 mg/mL; 5 min), and then rinsed with buffer for ∼1 h. The fluorogenic AP substrate, 4-MUP (200 µM), then was introduced using a syringe at a flow of ∼15 nL/s (6 mm/s). Scale bar, 25 µm. Inset: An intensity line plot that transects the particle.
maintaining a microparticle in the optical trap, in general we have found this step to be unnecessary, as particles can be transported to cellular targets without significant changes in size. We have explored the ability of these microparticles to interact with cells in real time. After fabrication in cell culture, a crosslinked protein particle was moved to a target cell and translocated through the plasma membrane into the cell cytosol (Figure 3A). Movement of the particle through the membrane was facilitated by the high power of the tweezers and exposure of the cells to hypotonic media, a procedure that reduces its resistance to puncture. Confocal sectioning on a different cell demonstrates the presence of a cross-linked BSA microparticle that was translocated into the cytosol of a cultured neuroblastoma-glioma cell using this procedure (Figure 3B). The ability to create and manipulate microparticles of this sort potentially has widespread cellular applications, ranging from highresolution stimulation of membrane-bound receptors to mediation of polarized intracellular signaling.12 Toward such goals, experiments were performed to evaluate the ability to catalyze chemical reactions using enzymatically active microparticles as catalytic “point” sources. A microparticle composed of avidin was fabricated inside a fused-silica capillary partially filled with photosensitizer/ protein solution by focusing the high-power optical trap through the capillary wall. Once fabricated, the avidin microparticle was transported using optical tweezers to the capillary wall, where it
became adherent and remained for the duration of the experiment. Cross-linked avidin retains significant biotin-binding capacity, providing a direct means for immobilizing a diverse range of biologically active or analytically useful compounds, including enzymes, antibodies, dyes, and reporter molecules. Here, the microparticle was decorated with biotin-AP; 4-MUP then was introduced into the capillary and fluorescence from enzymatically generated 4-MU was collected. The fluorescence intensity surface plot shown in Figure 3C was acquired during flow of the fluorogenic dye though the channel, demonstrating fluorescence from 4-MU generated specifically at the microparticle as well as background from nonspecific protein adsorption to the capillary walls. A dip can be seen in the middle of the particle fluorescence line scan (Figure 3C, inset), an effect that may result from limited diffusion of substrate/product to the interior of the microparticle. In Situ Chemical Gradients. The ability to selectively expose a cell or a small population of cells with a specific chemical effector would have broad utility in developmental and functional neuroscience, particularly if dosing could be achieved with arbitrary spatial control. To explore the potential applicability of biocompatible microfabrication to this goal, we have performed proofof-concept studies in which a real-time gradient of fluorescent molecules 4-MU is created through localization of a biotinylated enzyme to cross-linked avidin structures. Here, a structure (50 µm × 30 µm × 1 µm) composed of avidin was fabricated in a field of NG108-15 cells and functionalized by treatment with 0.1 mg/mL biotin-AP for 2 min (Figure 4A). The structure and cell field then were enclosed in a PDMS microchannel (cross-sectional area, 0.6 mm2), and a stream of 4-MUP was introduced at a flow rate of ∼0.5 mL/min (∼14 mm/s average flow rate). As shown in Figure 4B,C, 4-MUP was converted to fluorescent 4-MU principally at the structure, which was carried by the flow in a stream approximately the diameter of the structure for hundreds of micrometers. To characterize such product plumes under controlled conditions, a similar protein structure was fabricated in the absence of cells, and multiphoton fluorescence point measurements using a 100× objective were performed both downstream and upstream of the structure at varying heights. For the flow conditions and reagent concentration (200 µM 4-MUP) used in these studies, the concentration of fluorescent product immediately adjacent to a structure was ∼10-fold greater within the plume (32 ( 7 µM) than on the upstream side of the structure. Similarly low product levels were measured 10-20 µm both above and lateral to the plume so long as moderate flow rates were maintained. As evidenced by the downstream cell in Figure 4B (asterisk) that has accumulated significant 4-MU product, this approach can be used to differentially dose desired cells within unpatterned cultures. (Other cells in the shown field also have developed cytosolic fluorescence at somewhat lower levels, likely a result of nonspecific ecto-phosphatase activity.) The use of biocompatible microfabrication to localize enzyme reactors within cellular environments should be readily extendable to biologically relevant systems. For example, various peptides that mediate effects on neuronal differentiation and activity34,35 could be introduced into culture in a precursor form and activated by an immobilized phosphatase or protease. Moreover, although relatively large (34) Cazillis M.; et al. Eur. J. Neurosci. 2004 19, 798-808.
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
5093
Figure 5. Capillary nanoreactor in which GDH is photo-cross-linked near the outlet of a fused-silica capillary. (A) DIC image showing a series of GDH-BSA lines containing ∼105 molecular copies of GDH. Capillary inner diameter, 50 µm. (B) Enzymatic reaction scheme in which solution-phase GPT catalyzes the recycling of L-glutamate (Lglu) from R-ketoglutarate (R-KG), in the process converting L-alanine (L-ala) to pyruvate (pyr). Note that 1 equiv of NADH is generated each time glutamate is cycled through this pathway. (C) Lineweaver-Burk plot demonstrating scaling of reaction velocity. Abscissa and ordinate are plotted in units of nM-1 and (percent increase)-1, respectively.
Figure 4. Generation of chemical gradients within cellular cultures. (A) Transmission image (10× objective with a 1.5 times magnifier) showing a field of cultured NG108-15 cells after fabrication of a photocross-linked avidin scaffold (arrow) decorated with biotin-AP. (B) Fluorescence image of the same field subjected to a directional flow of 4-MUP (arrow denotes flow path). Fluorescent 4-MU product can be seen extending from the scaffold along the flow path and bathing a cell (asterisk) that develops high fluorescence due to dye uptake. (C) A surface plot showing fluorescence intensities from the central portion of the image in part B (enclosed within the rectangular box). Arrow denotes flow direction. In addition to the gradient extending from the scaffold, emission can be seen from two cells behind, and one directly bordering, the protein matrix.
structures were fabricated in the current proof-of-concept studies for ease of visualization, submicrometer spatial resolutionss suitable for subcellular dosingsshould be feasible using multiphoton fabrication. On-Column Nanoreactors. In addition to fabricating enzymatically active structures from avidin and biotin-conjugated proteins, direct cross-linking can be used to precisely pattern enzymes.36 Here GDH, an enzyme involved in processing the principal excitatory neurotransmitter in the central nervous system, glutamate, was used to create localized on-column reactors (Figure 5A), a format applicable to detection of components fractionated from complex biological matrixes. GDH catalyzes the oxidation of glutamate to R-ketoglutarate with the corresponding reduction of nonfluorescent NAD+ to NADH, a weakly fluorescent molecule. Because the equilibrium constant (Keq ≈ 10-15) for this reaction is highly unfavorable,37 a second enzyme, GPT, was included in solution to recycle glutamate (35) Lucius, R.; Gallinat, S.; Busche, S.; Rosenstiel, P.; Unger, T. Cell. Mol. Life Sci. 1999, 56, 1008-1019. (36) Basu, S.; Campagnola, P. J. Biomacromolecules 2004, 5, 572-579. (37) Colman, R. Glutamate Dehydrogenase (bovine liver); CRC Press: New York, 1991.
5094 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
(Figure 5B). Dual-enzyme schemes such as this have been used to increase the production of NADH for capillary electrophoresis assays and in immobilized enzyme sensors.38,39 A series of cross-linked enzyme matrixes was fabricated in contact with a fused-silica wall over a ∼125-µm segment of the capillary (Figure 5A), cut to size, positioned on an inverted microscope, and washed overnight with 100 mM borate (pH 9.5) to remove nonspecifically adsorbed enzyme. Similar enzyme structures were found to contain ∼5-fold greater protein concentration than the protein solution used in fabrication, yielding an estimate of GDH concentration in the structure of ∼50 µg/mL (150 nM). Reagent plugs (several millimeters long) containing NAD+, GPT, glutamate, and alanine were introduced into the capillary using hydrodynamic flow; reaction progress was monitored via continuous point measurements using a Ti:S laser focused adjacent to the enzyme matrix to excite fluorescence from NADH diffusing from the structure into solution. For 300 nM glutamate (a concentration approaching saturation), NADH increased to low micromolar concentrations adjacent to the structure after 10 min, indicating that femtomole quantities of NADH were generated by the reactor over this period. The entire GDH matrix was estimated to occupy ∼1 pL and to contain ∼105 molecular copies of GDH, yielding a molecular turnover rate for cross-linked enzyme of ∼102 s-1. (It is assumed that the rate of NADH generation was not limited by glutamate recycling, as the transaminase, GPT, was present in large molar excess and has been reported to have a turnover rate of ∼400 s-1.40) This value agrees well with the turnover rate (∼200 s-1) previously reported for solution-phase bovine liver GDH.41 A semiquantitative analysis was performed for glutamate concentrations that approach saturation of this reactor, demonstrating expected Lineweaver-Burke-like scaling (Figure 5C). (38) Wang, Z.; Yeung, E. S. J. Chromatogr., B: Biomed. Sci. Appl. 1997, 695, 59-65. (39) Pantano, P.; Kuhr, W. G. Anal. Chem. 1993, 65, 623-630.
To examine the extent to which nonspecific adsorption of GDH produced background in this system, reactions were attempted in control capillary tips that had been treated with photofabrication solution (BSA/GDH/FAD) but that had been rinsed of this solution without fabricating matrixes. Fluorescence was undetectable in these capillaries over a broad range of substrate concentrations and laser detection powers, indicating that nonspecific adsorption did not contribute to NADH production. Other groups previously have developed on-column enzymebased reactors for microfluidics, including photopolymerized solgel scaffolds and patterns created using surface photoattachment chemistries.42,43 To our knowledge, the current work represents the first demonstration of three-dimensionally controlled fabrication within microfluidic structures, an approach that promises to combine several of the strengths of earlier methods. Similar to on-column sol-gel formation, multiphoton cross-linking can create matrixes that contain high concentrations of protein and extend many micrometers from a surfacesthus offering the potential for efficient enzyme-substrate interaction, even under moderately high-flow-rate conditions in large cross-section channels. Like recent photoattachment procedures, multiphoton cross-linking is rapid and offers high spatial control over the deposition process. CONCLUSION At present, the principal limitation in these studies is nonspecific protein adsorption, the result of exposing unprotected glass and fused-silica surfaces to very high protein concentrations (typically at least 100 mg/mL) during the photo-cross-linking process. To avoid prohibitive nonlocalized enzyme activity throughout a fabrication chamber, extensive washing procedures were required, in some cases causing the release of adherent cells from surfaces. Even with such treatment, residual enzyme levels in some cases can generate significant steady-state product concentrations, particularly under stagnant flow conditions. Fortunately, a wide range of passivating agents (e.g., fibrinogen41) and surface coatings can be pursued to minimize adsorption. Alternative approaches, such as the photoincorporation of biotin into protein matrixes as they are created, may be useful for fabricating (40) De Rosa, G.; Burk, T. L.; Swick, R. W. Biochim. Biophys. Acta 1979 567, 116-124. (41) Rajas, F.; Rousset, B. Biochem. J. 1993, 295, 447-455. (42) Kato, M.; Sakai-Kato, K.; Jin, H.; Kubota, K.; Miyano, H.; Toyo’oka, T.; Dulay, M. T.; Zare, R. N. Anal. Chem. 2004, 76, 1896-1902. (43) Holden, M. A.; Jung, S.-Y.; Cremer, P. S. Anal. Chem. 2004, 76, 18381843. (44) Chiem, N. H.; Harrison, D. J. Clin. Chem. 1998, 44, 591-598. (45) Chiem, N.; Harrison, D. J. Anal. Chem. 1997, 69, 373-378. (46) Seong, G. H.; Heo, J.; Crooks, R. M. Anal. Chem. 2003, 75, 3161-3167.
structures with high avidin-binding capacities while reducing background associated with nonspecific protein adsorption. Fabrication of catalytically active protein matrixes provides a unique tool to tailor microenvironments for cellular and microanalysis applications. This approach offers the ability to create a broad range of topographical features (e.g., cables, surface adherent lines, pads, movable microparticles) and to achieve chemical functionality either through avidin-biotin interactions or via direct incorporation of effectors into matrixes. The avidinbiotin approach may have a distinct advantage under some circumstances, as it avoids questions of possible modification/ disruption of enzyme activity during photo-cross-linking and offers the potential to immobilize effector and reporter molecules that may not be amenable to photo-cross-linking. The approaches explored in this work should provide new capabilities for studying a variety of cell functions, ranging from single-cell biochemistry and development to perturbation and analysis of small populations of cultured cells. The threedimensional control inherent to multiphoton cross-linking should enable enzyme-based sensors to be fabricated within cellular environments in topographies that restrict diffusion of analytes from their cellular release sites, thus increasing the efficiency of mass-limited cellular analyses. Beyond its application to fundamental questions in cell biology, applications for this technology are likely to be found in regenerative medicine where precise chemical control of cell proliferation and differentiation is required. Recent innovations in microfluidics promise to revolutionize the biosciences by providing new approaches to diagnostics,44,45 enzymatic assays,46 and cell biology; 10,11 on-demand, high-resolution microfabrication complements and adds versatility to these technologies by providing a means to alter chemical and mechanical environments after fabrication of initial microfluidic architectures. ACKNOWLEDGMENT We thank A. Bardo at the Microscopy and Imaging Facility, Institute for Cellular and Molecular Biology (ICMB), University of Texas, for assistance with confocal imaging, R. Hill and B. Kaehr for contributing to SEM imaging, and J. Lyon for assistance with AFM imaging. J.B.S. is a fellow at the ICMB. Funding from the Robert A. Welch Foundation (Grant F-1331), the National Science Foundation (Grant 0317032), and the National Institutes of Health (Grant 8R21EB000423-02) is gratefully acknowledged. Received for review May 6, 2005. Accepted June 24, 2005. AC0507892
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
5095
