|Home | About | Journals | Submit | Contact Us | Français|
Control of the spatial arrangement of proteins on surfaces is essential in a number of emerging technologies, including protein microarrays, biosensors,1 tissue engineering, and regenerative medicine.2 Patterning is also a powerful tool in cell biology, wherein cell arrays are used to elucidate the factors that mediate migration, proliferation, and cell-cell interactions.3 Although photolithography holds a preeminent place as a patterning method in the microelectronics industry, optical lithography of proteins has been hampered by the need either to use traditional chemical photoresists or to modify proteins chemically by attachment of photoreactive functional groups; both methods can compromise protein function.4
Production of a protein “photoresist” without the need for post-translational chemical modification would require an intrinsically photoreactive protein. Recently, the incorporation of photo-reactive, non-canonical amino acids into proteins via site-specific5 and residue-specific techniques has been reported.6 Here we describe the microbial expression of artificial proteins bearing the photosensitive non-canonical amino acid para-azidophenylalanine (pN3Phe). The recombinant proteins, designated artificial extracellular matrix proteins with aryl azides (aECM-N3), belong to a family of engineered proteins designed to exhibit mechanical properties similar to those of native elastins7 and to support adhesion of mammalian cells through cell-binding domains (CS5 or RGD) derived from fibronectin (Fig 1A).8 These proteins can be crosslinked efficiently upon irradiation at 365 nm. The physical properties of the crosslinked films can be controlled by changing the pN3Phe content, and thin films can be patterned on surfaces via photolithographic techniques. We demonstrate the utility of the method by creating cell arrays through selective cell attachment to photolithographically prepared protein patterns.
aECM-N3 variants were expressed in Escherichia coli cultures supplemented with pN3Phe (Supporting Information). Incorporation of pN3Phe into the recombinant proteins relies on activation of the photosensitive amino acid by the phenylalanyl-tRNA synthetase (PheRS) of the bacterial expression host. The PheRS used for this study was a previously characterized mutant with relaxed substrate specificity.9 This method results in statistical decoding of phenylalanine (Phe) codons placed at regular intervals in the coding sequence.9 Proteins were expressed in a Phe-auxotrophic E. coli strain and purified by exploiting the temperature-dependent phase behavior of proteins that contain elastin-like repeats.10 Incorporation efficiency was determined by integration of the aromatic proton signals in the 1H NMR spectra of the purified proteins; the extent of Phe replacement varied from 13% to 53%, depending on the concentration of pN3Phe in the expression medium (Supporting Information).
Understanding the response of the photoreactive protein to irradiation is crucial for high-resolution pattern formation. We measured the rate of azide decomposition under irradiation by monitoring loss of the characteristic infrared asymmetric stretch at 2130 cm−1 (Fig. 1B).11 Measurements were performed on thin films of CS5-N3 spin-coated directly onto zinc selenide wafers and irradiated using a Karl Suss contact aligner filtered to 365 nm in constant intensity (7 mW/cm2) mode, with a quartz wafer in place of the mask. Azide loss under these conditions was rapid, following first-order kinetics with a half-life of 34 seconds (Fig. 1C). It is noteworthy that none of the other infrared bands was noticeably altered, indicating that irradiation under the conditions used here activates the aryl azide without substantial modification of the other canonical amino acids. Aryl azides have been used previously to effect photochemical crosslinking in protein and nucleic acid systems.12
Elastic moduli of irradiated CS5-N3 films were determined by uniaxial tensile testing under simulated physiological conditions (Fig. 1D). As expected, the elastic modulus correlated with the extent of pN3Phe incorporation. Irradiated CS5-N3 films in which 30, 41, or 53% of the encoded Phe residues were replaced by pN3Phe yielded elastic moduli of 0.53 ± 0.10, 0.94 ± 0.09, and 1.39 ± 0.09 MPa, respectively, values near the range characteristic of elastins (0.3 – 0.6 MPa).7 Replacement of less than 20% of the encoded Phe residues produced films too weak to test, and films made without pN3Phe yielded no evidence of crosslinking. The capacity to vary the modulus by altering the pN3Phe concentration in the expression medium is an attractive feature of this method, as recent work has highlighted the role of mechanical transduction mechanisms in mediating the physiology of adherent cells.13
To investigate the potential of photoreactive proteins as substrates for studies of cell adhesion and growth, we created patterns of adherent fibroblasts on proteins patterned by photolithography. Protein films created by spin coating 12.5 mg/mL solutions of RGD-N3 in dimethylsulfoxide directly on poly(ethylene oxide) (PEO)-coated glass slides were clear and homogeneous by optical microscopy. Protein films were irradiated for 60 seconds at 365 nm through a chrome-on-quartz mask. Stripping of the masked areas was accomplished by washing in 6 M guanidine hydrochloride.
Fluorescence immunolabeling with an anti-T7 tag antibody showed that the protein was localized only within the irradiated areas of the pattern. Films prepared from protein lacking pN3Phe formed no detectable patterns even after prolonged exposure. Non-contact atomic force microscopy (AFM) of RGD-N3 patterns revealed uniform protein films. Films spun at 1000 rpm were 467 or 750 nm thick when imaged dry or hydrated, respectively. (Supporting Information).
To create cell arrays, Rat-1 fibroblasts were deposited on RGD-N3 patterns in the absence of serum. After 4 hours of incubation, the unattached cells were removed by mild washing with phosphate buffered saline (PBS) revealing cell patterns (Fig. 2). Cell monolayers in the interior of the protein regions were indistinguishable from monolayers grown on fibronectin coatings; however, cells positioned along the RGD-N3 pattern edges were elongated parallel to the pattern border, consistent with previous studies (Supporting Information).14
To determine whether cells specifically recognize the RGD cell-binding domain, we compared cell spreading on uniformly photocrosslinked RGD-N3 and RDG-N3 (sequence-altered, negative control) films (Fig. 3). After 4 hours of incubation, Rat-1 cells spread well on RGD-N3 films, although the extent of spreading was reduced in comparison to that observed on the fibronectin control. In contrast, cells did not spread on RDG-N3 and resembled cells plated on bovine serum albumin (BSA).
The availability of intrinsically photoreactive proteins enables new approaches to protein patterning. The technical simplicity of the method allows rapid production of samples with a wide variety of feature shapes and sizes, while permitting straightforward engineering of the elastic modulus of the crosslinked protein. The method is a promising approach to the study of adherent cells, providing control over mechanical properties, ligand-receptor interactions, and geometric shape. Applications in medical devices, tissue engineering, and array technologies are readily imagined.
Methods for cloning, protein expression and purification, patterning, and cell studies. AFM and phase contrast images are also provided.
We thank Michael Diehl, Alireza Ghaffari and Nandita Sharma for helpful discussion, and Kechun Zhang for help in preparing the patterned substrates. Supported by the NSF Center for the Science and Engineering of Materials at Caltech, NIH GM 62523 and EB 01971, an NIH predoctoral fellowship to SAM, a Whitaker Graduate Fellowship to JCL, and the Joseph J. Jacobs Institute for Molecular Engineering for Medicine.