Gradient lithography of engineered proteins to fabricate 2D and 3D cell culture microenvironments
Spatial patterning of proteins is a valuable technique for many biological applications and is the prevailing tool for defining microenvironments for cells in culture, a required procedure in developmental biology and tissue engineering research. However, it is still challenging to achieve protein patterns that closely mimic native microenvironments, such as gradient protein distributions with desirable mechanical properties. By combining projection dynamic mask lithography and protein engineering with non-canonical photosensitive amino acids, we demonstrate a simple, scalable strategy to fabricate any user-defined 2D or 3D stable gradient pattern with complex geometries from an artificial extracellular matrix (aECM) protein. We show that the elastic modulus and chemical nature of the gradient profile are biocompatible and allow useful applications in cell biological research.
KeywordsProtein patterning Protein lithography Protein gradient DMD (digital micromirror device) Protein engineering Non-canonical amino acids
Protein patterning is becoming increasingly important for a myriad of applications in biosensors (Ekins 1989; Wang et al. 2007a; Zhu and Snyder 2003), developmental biology, stem cell biology, and tissue engineering (Fischbach et al. 2007; Herbert et al. 1997; Khademhosseini et al. 2006; Shen et al. 2008; Von Philipsborn et al. 2006). It is generally accepted that chemical and molecular cues play a critical role in cell differentiation and migration during early embryonic development (Gurdon and Bourillot 2001). However, applying this property for tissue engineering applications has been thus far limited, primarily due to the technical constraints in producing gradients of molecular cues or cell binding ligands as close as native extracellular matrix environments. On the other hand, it is also generally acknowledged that other factors including the topographic and mechanical properties of the environment can affect cell development (Clark et al. 1991; Engler et al. 2006; Janmey and McCulloch 2007). As an example, both molecular (Nishiyama et al. 2003) and topographic (Clark et al. 1991) cues can guide axon growth in tandem. It is difficult to identify specific contributions of each factor during certain cell developmental stages due to technical limitations in reproducing the effects with comprehensive microenvironment control ex vivo.
While many protein patterning techniques, including contact imprinting (Choi and Newby 2003; Von Philipsborn et al. 2006), diffusion processes (Dertinger et al. 2002; Georgescu et al. 2008; Jeon et al. 2000; Luo and Shoichet 2004; Wang et al. 2008), and optical lithography (Herbert et al. 1997; Hypolite et al 1997; Li et al. 2005) are available, each of these methods has technical limitations that restrict its usefulness for biological research. Close approximations of in vivo biological environments require that certain characteristics of protein patterns be considered. First, in order to study cellular processes that occur over days or weeks, stable protein patterns created using covalent binding is preferred over physical adsorption methods (Zhu and Snyder 2003). Second, molecular cues or cell binding ligands should be presented as continuous gradients that are convenient to design and reconfigure. Many contact imprinting methods only produce discrete dots that can be patterned to form a gradient density (Von Philipsborn et al. 2006). Diffusion processes, e.g. integrated microfluidic channels (Dertinger et al. 2002; Georgescu et al. 2008; Jeon et al. 2000; Wang et al. 2008), are able to produce continuous gradient protein patterns, however they often require new microdevice design and fabrication to reconfigure the gradient profile. Third, as mentioned above, mechanical and topographic properties of the cell culture environments also play very important roles in cell development (Clark et al. 1991; Engler et al. 2006; Janmey and McCulloch 2007, Mai et al. 2007). Therefore, a flexible protein patterning method is desired; one that can be used on traditional cell culture substrates (e.g., glass and tissue-culture polystyrene) and that has mechanical properties similar to the native ECM environment. The optimal method would be highly efficient to enable high throughput patterning and screening of multiple microenvironments. While scanning optical lithography is a very successful way to produce protein gradients in polymer environments, (Herbert et al. 1997; Hypolite et al 1997; Li et al. 2005) it is hampered by its relative slowness. Finally, cells in vivo grow, migrate, and differentiate in three-dimensional (3D) microenvironments, however many in vitro protein patterning methods are limited to patterning of two-dimensional (2D) substrates. In this paper, we present a novel protein patterning method that addresses each of the limitations described above. We demonstrate a simple dynamic mask lithography method to fabricate any user-defined 2D or 3D gradient pattern from an artificial extracellular matrix (aECM) protein. The cell-binding domains are covalently incorporated into the aECM environments and produced in continuous gradient patterns. This is a high throughput optical lithography method where the pattern shapes and gradient profiles can be conveniently reconfigured. We test the generated patterns and show them to be compatible for cell cultures.
In order to develop a flexible protocol capable of fabricating both 2D and 3D patterns of arbitrary geometry, we employed a dynamic mask projection lithography system that has been designed previously for microstereolithography applications (Lu et al. 2006; Sun et al. 2005). This system utilizes a Digital Micromirror Device (DMDTM, Texas Instruments) as a dynamic lithography mask instead of a conventional, permanent mask or other scanning methods. This lithography process offers several distinct advantages over conventional masks or scanning protocols including: parallel patterning, high throughput prototype development, and particularly easy procedures for obtaining any desired gradient via gray scale lithography. Another major advantage of this system is the high resolution, 3D complex structures that can be fabricated (Lu et al. 2006; Sun et al. 2005; Zhang et al. 1999). Due to the parallel processing capabilities of DMD, the applications have been extended to parallel manipulating, sorting, and stimulating of cells (Chiou et al 2005; Wang et al. 2007b). However previous reports regarding the applications of this fabrication tool have been primarily limited to photo-sensitive, synthetic polymer materials. Because proteins are not intrinsically photoactive and are often denatured or degraded by the reagents commonly used in photolithographic processing, protein patterning using dynamic mask lithography has not been previously described.
Recent reports have demonstrated that a genetically modified strain of Escherichia coli is capable of synthesizing intrinsically photoactive proteins that can be directly patterned through optical lithography (Carrico et al. 2007). This technique has been used to synthesize a family of artificial extracellular matrix proteins (designated as aECM-N3) with partial replacement of the canonical amino acid phenylalanine with the photosensitive non-canonical amino acid para-azidophenylalanine (pN3Phe) (Carrico et al. 2007). Multiple cell-binding domains, such as the CS5 or RGD sequences derived from fibronectin (Heilshorn et al. 2005, 2003; Liu et al. 2004) can be designed and inserted into the aECM-N3 proteins to promote cell adhesion through specific cell-surface integrin receptors. Additionally, the elastic moduli of the photocrosslinked aECM-N3 protein film can be modulated through the ratio of pN3Phe incorporated and through the intensity and duration of light exposure. Taken together, these results demonstrate that aECM-N3 photocrosslinked scaffolds have similar mechanical and cell binding properties to native extracellular matrix proteins, making them ideal materials for potential use in medical devices, drug and cell delivery vehicles, and tissue engineering scaffolds (Carrico et al. 2007).
By combining dynamic mask projection lithography and protein engineering with non-canonical photosensitive amino acids, we are able to create both 2D and 3D protein patterns with cellular-level resolution using engineered aECM-N3 proteins.
2 Materials and methods
2.1 aECM-N3 synthesis and purification
2.2 aECM protein lithography
The principle of the gradient lithography of aECM-N3 proteins is shown in Fig. 1b. For 2D lithography, the aECM-N3 proteins were dissolved in water (100 μg/μl, 4°C) and spin-coated at 1,000 rpm onto an aminated glass coverslip at 4°C. Clean glass coverslips were prepared by sonication in ethanol saturated with potassium hydroxide for 15 min. The slides were then washed with water and aminated in a solution of ethanol-acetic acid-diethyldiaminotriethoxysilane (93:5:2) for 15 min. Aminated surfaces were washed thoroughly with deionized water and dried under a stream of argon. For 3D lithography, the aECM-N3 proteins were dissolved in dimethyl sulfoxide (DMSO, 100 μg/μl, RT) and pipetted into the center of a glass coverslip (20 μl on an 18-mm round coverslip), surrounded by a thin circle of vacuum grease, and sandwiched with a rectangular coverslip (24 × 36 mm) to a separation thickness of 2 mm. The vacuum grease applied at the edges provided a good seal to prevent evaporation.
Details of the DMD projection lithography system have been described previously (Sun et al. 2005; Wang et al. 2007b). After irradiation, samples were washed three times by agitation in 0.1% Sodium Dodecyl Sulfate (SDS) for a minimum of 4 h, thoroughly rinsed with water, and stored in PBS at 4°C until further use.
2.3 aECM protein pattern staining and characterization
To visualize the 2D protein patterns, the T7 tags at the N-termini of the engineered proteins were specifically labeled using fluorescent antibodies. Briefly, samples were blocked with 1 ml 10% bovine serum albumin in PBS for 30 min followed by addition of 0.5 μl T7 tag-primary antibody and 6 h incubation at room temp. Samples were rinsed three times with PBS for 5 min, then blocked with 1 ml 10% bovine serum albumin in PBS for 30 min. Secondary antibody (100 μl of anti-mouse-Cy2 reconstituted to 1 mg/ml with water and then diluted with an equal volume of glycerol) was added and incubated for 2 h, samples were rinsed four times with PBS, 5 min each. Then the patterns were imaged using a confocal microscope.
For characterization of the 3D lithography protein patterns, the hydrated protein patterns were imaged in dark field by an inverted optical microscope (TE2000, Nikon). Then the sample was dried and its topographic profile was measured by the profilometer (Alpha-Step IQ Surface Profiler). After 20 nm gold was coated through sputtering, the sample was characterized by SEM (quanta 200 HV from FEI Company).
In order to get the elastic moduli of crosslinked aECM-N3 protein, a 1 mm thick layer of crosslinked protein film was made and characterized in PBS solution by AFM (Dimension 3,100, Vecco). The standard AFM force curve measurement was performed on the full hydrated protein sample with a relative blunt tip (cantilever spring constant 0.6N/m, NSC19/Cr-Au, MikroMasch). The elastic moduli were estimated through the slope of the AFM force curve (Dimitriadis et al. 2002).
2.4 PC-12 cell culture and staining
PC-12 cells (ATCC, Manassas, VA) were cultured in PC-12 complete growth media (F12 Kaighn’s media, 10% horse serum, 5% fetal bovine serum, and 1% penicillin-streptomycin) at 37°C, 5% CO2. Cells were differentiated in PC-12 differentiation media (F12 Kaighn’s media, 1% penicillin-streptomycin, 50 ng/mL recombinant human β-NGF (R&D Systems, Minneapolis, MN)) at 37°C, 5% CO2. Media was changed every other day for all cell experiments. aECM-N3 patterned substrates were soaked in poly-L-lysine for 1 h to promote cell adhesion. PC-12 cells were seeded directly on the 3D patterned substrate and cultured for 48 h in complete growth media followed by one week of culture in differentiation medium. Cultures were then washed 3× with PBS buffer, stained with 1 μg/ml DAPI (Roche) and immunostained with neuronal class III β-tubulin (Tuj1) primary rabbit monoclonal antibody (1:250, Covance, Berkeley, CA) followed by secondary antibody (goat-anti-rabbit) conjugated with Alexa Fluor 488 (1:500, Invitrogen, Carlsbad, CA). Immunostained samples were fluorescently imaged with an inverted Zeiss Axiovert 200 microscope (Oberkochen, Germany, 20× objective) and a CCD camera.
3 Results and discussion
3.1 2D concentration gradient protein pattern fabrication and characterization
3.2 3D gradient protein pattern fabrication and characterization
3.3 Biocompatibility demonstration with PC-12 cell culture
We have developed a tool that is highly efficient in making user-defined spatial 2D and 3D gradient protein patterns. We achieve this by combining recent advances in synthesizing intrinsically photoactive proteins with a projection lithography system that employs a Digital Micromirror Device (DMD) as a dynamic mask. The simplicity of our technique makes it highly amenable to high throughput applications requiring such gradients.
The system that we report in this paper is highly flexible. While we have demonstrated the viability of our system using a few specific user-defined complex shapes, we believe that the dynamic mask being utilized here makes reconfiguration to any desired gradient profile quite easy. The efficiency and ease of use makes it particular useful for complex shape and gradient profile patterning compared to currently available techniques. One prevailing technique to create gradient patterns uses integrated microfluidic channels with diffusion profiles. Gradient profiles created by this technique are limited by the diffusion process—therefore, redesign and reconfiguration of such profiles can be complicated.
The properties of the pattern obtained using our methods are very suitable for high throughput biological studies. Pattern size and resolution can be varied by delivering light through different reduction rate lenses. Such flexibility in pattern size and resolution provides the opportunity for studies ranging from subcellular to tissue-length scales. For example, a millimeter-sized pattern yields submicron resolution suitable for subcellular microenvironment control (Takayama et al. 2001). In contrast, patterned areas up to centimeters can be easily produced for studying cell interactions, migration, and organization within groups of cells. As proof of the utility of this technique, we demonstrate that the aECM-N3 patterned substrates are biocompatible. PC-12 cells grown on these substrates demonstrate typical neuronal morphologies with neuritic processes observed to exhibit path-finding modulated by 3D gradient topography.
The fabricated patterns are very stable due to covalent crosslinks between the protein polymers and the aminated glass substrate. The mechanical properties of the aECM can be tuned by tailoring the concentration of N3 incorporated, the extent of crosslinking, and the weight percentage of the protein solution. In addition, due to the templated synthesis of aECM engineered proteins, the scaffolds can be easily tailored by including multiple peptides with distinct functionalities for specific cell and tissue applications (Carrico et al. 2007; Straley and Heilshorn 2009). Additionally, this method can be extended to create patterns of multiple proteins by following the initial patterning step with subsequent patterning steps. These types of complicated gradient patterns have been inaccessible to experimentalists to date due to technical barriers.
The capability of fabricating complex 3D microstructures makes this technique ideal for the creation of microenvironments with highly tailored chemical, mechanical, and topographical patterning. The ability to easily create these patterns across a range of length scale – from subcellular to tissue-level – makes this technique ideal for studying cell-environment and cell-cell interactions. Thus we have developed a method that can design and fabricate a wide variety of 2D and 3D engineered in vitro cellular environments for future research in developmental biology, regenerative medicine, and tissue engineering.
Phenylalanyl-auxotrophic Escherichia coli harboring the aECM plasmid and mutant PheRS gene was a kind gift from David Tirrell, Caltech. S. W., A. W. and X. Z. acknowledge financial support from NIH Nanomedicine Development Center (Center for Cell Control, PN2 EY018228) and NSF Nano-scale Science and Engineering Center (NSEC, DMI-0327077).
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.