A Controlled Platform for Studying Cell–Ligand Interactions - American

Aug 17, 2012 - The gold islands were functionalized with alkanethiol self- ... Gold island arrays are defined in the form of a matrix ..... (1) Love, ...
0 downloads 0 Views 1MB Size
Letter pubs.acs.org/NanoLett

Mixed Alkanethiol Monolayers on Submicrometric Gold Patterns: A Controlled Platform for Studying Cell−Ligand Interactions Rami Fishler,†,‡ Arbel Artzy-Schnirman,†,‡,§ Elad Peer,†,‡ Ron Wolchinsky,§ Reuven Brener,∥ Tova Waks,⊥ Zelig Eshhar,⊥ Yoram Reiter,‡,§ and Uri Sivan*,†,‡,∥ †

Department of Physics, ‡The Russell Berrie Nanotechnology Institute, §Department of Biology, and ∥Solid State Institute, TechnionIsrael Institute of Technology, Haifa 32000, Israel ⊥ Department of Immunology, Weizmann Institute of Science, Rehovot 76100, Israel S Supporting Information *

ABSTRACT: Nanoscale organization of surface ligands often has a critical effect on cell−surface interactions. We have developed an experimental system that allows a high degree of control over the 2-D spatial distribution of ligands. As a proof of concept, we used the developed system to study how T-cell activation is independently affected by antigen density and antigen amount per cell. Arrays of submicrometer gold islands at varying surface coverage were defined on silicon by electron beam lithography (EBL). The gold islands were functionalized with alkanethiol self-assembled monolayers (SAMs) containing a small antigen, 2,4,6-trinotrophenyl (TNP), at various densities. Genetically engineered T-cell hybridomas expressing TNP-specific chimeric T-cell antigen receptor (CAR) were cultured on the SAMs, and their activation was assessed by IL-2 secretion and CD69 expression. It was found that, at constant antigen density, activation increased monotonically with the amount of antigen, while at constant antigen amount activation was maximal at an intermediate antigen density, whose value was independent of the amount of antigen. KEYWORDS: T-cell activation, patterned antigen surface, immobilized antigen, cell interaction with SAM, X-ray photoelectron spectroscopy of SAMs

T

and amount on cell behavior. The platform consists an array of submicrometer gold islands defined by EBL at a desired surface coverage. The gold surface is then functionalized with a mixed SAM containing a ligand at a defined density controlled by the molecular ratio of ligand-containing molecules and diluting molecules in the mixed SAM. This platform is easy to manufacture using standard EBL and SAM deposition techniques, and it facilitates the investigation of cell interaction with virtually any surface ligand. Figure 1 illustrates a typical implementation of the described platform. Gold island arrays are defined in the form of a matrix with rows of varying ligand density and columns of varying gold coverage. The response of cells to the molecular pattern is assessed at either a fixed ligand density and varying ligand amount (surfaces with identical ligand densities and varying gold coverage), or a fixed ligand amount and a varying ligand density (surfaces with increasing ligand density and decreasing gold coverage). The size of gold islands and gaps between them is kept significantly smaller than the cell’s diameter (typically ∼10 μm). Consequently, the gold coverage controls the mean area of monolayer that interacts with each cell without affecting

he interaction of cells with surface ligands is fundamental to a wide range of cell activities such as differentiation, surface adhesion, and cell−cell signaling. Traditionally, these processes were studied in the context of ligand/receptor molecular interaction properties and amount of interacting molecules. It is now evident, however, that the detailed nanoscale organization of ligands often has a detrimental effect on the interaction outcome. Self-assembled monolayers (SAMs) comprise unimolecular layers of organic molecules, adsorbed on a solid surface.1 They are easily designed to display different biologically active molecules at a defined orientation and density and therefore represent an ideal platform for studying cell−surface interactions. For example, Houseman and Mrksich2 used mixed SAMs of two molecule types to control surface densities of cellanchoring peptides and study their effect on cell adhesion. Micro- and nanopatterning of SAMs provide means to control the spatial organization of molecules on the surface. Senaratne et al. used SAM functionalized gold arrays of 45 nm to 1 μm features to study the spatial regulation of cellular signaling in MAST cells.3 Arnold et al.4 used a close-packed template of gold nanodots functionalized with cyclic RGDFK peptide to examine how cell adhesion is affected by ligand density defined by the local ligand-to-ligand distance and overall amount of ligand available for interaction with each cell. Inspired by the work of Arnold et al.,4 we developed a simple platform that allows differentiating the effects of ligand density © 2012 American Chemical Society

Received: July 16, 2012 Revised: August 12, 2012 Published: August 17, 2012 4992

dx.doi.org/10.1021/nl302619p | Nano Lett. 2012, 12, 4992−4996

Nano Letters

Letter

The system comprises arrays of submicrometer gold islands defined on a silicon surface by EBL. The gold islands are functionalized with a mixed SAM deposited by immersion in an ethanol solution of HS-(CH2)11(CH2CH2O)4NH2 and HS(CH2)11(CH2CH2O)2OH. The NH2 groups are then modified with 2,4,6-trinitrophenyl (TNP) by reaction with 2,4,6trinitrobenzene sulfonic acid (TNBS), resulting in a TNP presenting SAM (Figure 2a). As a model system for cell activation and response to immobilized ligands we use CAR expressing T-cells, termed STB cells. These engineered TNPspecific T-cell hybridomas (Figure 2b) express CARs with an extracellular TNP binding domain and an intracellular signaling domain.11 When interacted with TNP-presenting surfaces, the molecular interaction between the chimeric TCR and its ligand, TNP, results in cell activation. The level of activation is assessed by IL-2 secretion or CD69 expression. In our experimental system, antigen density is controlled by the molecular composition of the SAM, namely, the ratio between TNP terminated molecules and hydroxyl terminated molecules, while the amount of molecules that interact with each T-body is determined by the gold coverage. We performed X-ray photoelectron spectroscopy (XPS) analysis to confirm that the produced SAMs are densely packed, sufficiently clean, and that the percentage of amine containing molecules on the surface, %NH2surface, is correlated with its percentage in the solution, %NH2solution (see Supporting Information (SI)). It is common to assume a linear relationship between the fraction of molecules in the solution and in the resulting SAM, especially when trends rather than precise densities are considered.12−14 Our analysis showed, however, that amine terminated molecules adsorb preferably to the gold at low %NH2solution as previously reported.15 We therefore calibrated the monolayer composition by XPS at higher %NH2solution concentrations where XPS was sensitive enough and extrapolated the data using a competitive coadsorption kinetic model16 to the experimentally relevant concentrations (%NH2solution ≈ 1%); see the SI for details. Better accuracy, when important, can be achieved by radio-labeling of the TNP carrying molecules in the monolayer.2 A similar analysis has been applied to the unpatterned SAMs used for the complementary experiments described below. To explore how antigen density and amount affect T-cell activation we fabricated by EBL 1 × 1 mm2 arrays of square gold islands with side dimensions varying between ∼320 nm and ∼740 nm at a ∼895 nm pitch. Surfaces were functionalized with TNP presenting SAMs of varying composition and tested for their ability to activate TNP specific T-bodies. Figure 3 shows activation of CAR expressing cells for nine levels of gold coverage and seven TNP densities. Activation was assessed by measuring secreted IL-2 concentration in the medium 48 h after activation using an enzyme-linked immunosorbent assay (ELISA). All measurements were carried out using cells from the same culture on the same day. The results for 100% gold coverage average two independent measurements on two different surfaces. The optimal %NH2surface for activation was found to be approximately 9%, practically independent of gold coverage (Figure 3a). To examine the effect of antigen density on activation at a constant amount of antigen per cell, it was useful to plot IL-2 secretion as a function of TNP density and gold coverage (Figure 3b). Approximate lines of constant antigen amount per cell are marked by dashed lines. It is evident that at a constant amount of antigen per cell there is a clear optimal

Figure 1. Concept of a typical experiment. The platform provides independent control over ligand density and amount. Along the blue horizontal lines ligand amount varies at a constant ligand density. Along the green diagonal lines ligand density changes at a constant amount per cell.

the amount of cells interacting with the monolayer as would be the case for macroscopic gold islands. As a proof of concept we apply our approach to test how Tcell activation is affected by density and amount of antigens that bind a specific chimeric T-cell receptor.5 This question is particularly intriguing since the role of antigen aggregation in the mechanism of T-cell activation is not yet clear.6−9 We have recently shown that molecularly engineered T-cell hybridomas can be activated by electroactive SAMs in a switchable manner.10 Figure 2 shows a variant of this system designed for the current purpose.

Figure 2. Experimental setup. (a) The modification reaction leading to the antigen monolayer. (b) Experimental system. 4993

dx.doi.org/10.1021/nl302619p | Nano Lett. 2012, 12, 4992−4996

Nano Letters

Letter

Figure 3. Activation of TNP-specific T-bodies by patterned gold island arrays. (a) IL-2 secretion as a function of antigen density for different levels of gold coverage. Error bars represent standard deviation of data points at each %NH2solution from the corresponding linear regression curve. (b) IL-2 secretion plotted as a function of antigen density and gold coverage. Dashed lines depict approximate lines of constant antigen amount per cell. IL-2 concentration for %NH2surface = 21.7% was taken as a baseline based on a similar experiment showing no detectable activation at this antigen concentration.

Figure 4. Activation of T-bodies on patterned surfaces. (a) Same data as in Figure 3. (b) Data from an independent experiment. Error bars represent standard deviation of data points at each %NH2solution with respect to the corresponding linear regression curve.

amount indicates no optimal antigen amount for cell activation. The decrease in activation can therefore be safely associated with the high antigen density regardless of its total amount per cell. To further characterize the activation of cells we have also used flow cytometry to measure CD69 expression, an early marker for T-cell activation. CAR expressing cells were seeded on TNP presenting SAMs deposited on unpatterned gold surfaces, and CD69 expression was measured 3 h after seeding. The same trend in activation was observed as for IL-2 secretion, further verifying the emerging conclusion (Figure S4 of the SI). Moreover, the enhancement of CD69 expression proves that cells are activated using the expected signal transduction path. Our results indicate an optimal antigen density for T-cell activation. To eliminate the possibility of such an optimum resulting from mutual steric hindrance of antigens we characterized the binding of a soluble anti-TNP antibody to TNP presenting SAMs of varying %NH2solution by an ELISA assay (Figure S5a, SI). In this experiment, we have used the low

antigen density for activation, and this optimum is independent of antigen amount per cell. Figure 4 shows the effect of antigen amount per cell at a constant antigen density. An approximate linear relation between cell activation and amount of antigen per cell at a constant antigen density is evident, indicating that antigen density and not its amount per cell is responsible for the decreased activation at high antigen densities. In other words, there exists an optimal antigen density for activation and no optimal antigen amount since at a constant antigen density, the more antigen interacts with each cell the higher the average activation. Consider for instance the red line in Figure 4a, corresponding to a higher density than the optimal one. If an optimal amount of antigen existed, it would have surely been reached along this line, and a decrease in activation would have been observed at high antigen amounts. However, the monotonic increase in activation with an increased antigen 4994

dx.doi.org/10.1021/nl302619p | Nano Lett. 2012, 12, 4992−4996

Nano Letters

Letter

affinity murine IgM SP6 antibody,17 the sequence of which was used by Eshhar et al.11 to synthesize the chimeric receptor of the T-bodies. In parallel to the binding experiment, we incubated CAR expressing cells on similar surfaces and measured IL-2 secretion after 24 h (Figure S5b). Although activation decreased pronouncedly at a %NH2surface of 9.5%, and almost no activation was observed at a %NH2surface of 14% (Figure S5b), we did not observe substantial change in antibody binding in all examined antigen densities (Figure S5a). This fact indicates that the receptor binds the surface at antigen densities higher than the optimal density for activation. Therefore, the decreased activation at high antigen densities could not be attributed to steric hindrance of receptor binding. It should be noted, though, that the binding of a soluble pentameric IgM antibody to the antigen differs in many aspects from the binding of a membrane associated single chain domain. To conclude, we developed a novel experimental method enabling independent control over surface density and amount of surface-immobilized ligands. We demonstrated its usefulness by exploring how T-cell activation is independently affected by antigen density and amount. Two main results emerged. (1) At a constant antigen density the probability for T-cell activation increases monotonically with the amount of antigen per T-cell. (2) At a constant amount of antigen per T-cell, activation reaches an optimum at an intermediate antigen density that is independent of antigen amount per T-cell. In addition, we showed evidence that the decreased activation at high antigen densities does not result from steric hindrance of receptor binding. These results concord with models that suggest a direct link between antigen density and signal initiation. However, it is not clear why high antigen density results in reduced activation. Artifacts associated with the potential toxicity of high NH2/ TNP concentration surfaces to cells were ruled out since cell viability counts showed no correlation with inhibition of activation. Moreover, CD69 expression showed the same trend of activation as IL-2 secretion although measured only 3 h after activation. This period was not sufficiently long for cell apoptosis, and therefore toxicity could not have caused the decrease in cell activation. The systematic variation of activation with TNP density and amount and the similarity between repeated experiments indicate that monolayer inhomogeneity, if existent, plays a minor role. We therefore attribute the decreased activation to the tightly packed configuration of chimeric receptors. This structure may impose geometrical constraints on phosphorylation of cytoplasmic signaling domains. Alternatively, the presence of closely packed surface bound TCRs at the periphery of the contact area may block TCR diffusion to the center of the synapse. The same mechanism may block the passage of membrane associated kinases such as Lck and fin that are necessary for activation. These explanations agree with the fact that the optimal antigen density was independent of the amount of antigen per cell. We chose engineered T-cells expressing CARs as a model for T-cell activation for two reasons. First, activation of CAR expressing T-cells is simpler than activation of T-cells and involves fewer components. Second, T-cells expressing CARs can be activated by small molecules, whose distribution on patterned surfaces is better controlled compared with large protein complexes. The downside of studying these cells is in their unknown relevance to the study of natural T-cell

activation. However, T-cells expressing CARs with specificity toward tumor-associated antigens have been introduced recently in clinical trials and demonstrated promising results in patients, inducing potent killing and elimination of target cells.18,19 These results suggest that the activation of CAR expressing engineered T-cells may resemble and exhibit similar properties to the activation and response of naturally occurring T-cells. The present experiment should therefore be considered a first step on the way to studying the effect of the same parameters on activation of normal T-cells. T-cells have been previously investigated by patterned surfaces containing either immobilized ligands20,21 or mobile ligands incorporated in the upper leaflet of a supported lipid bilayer.22,23 While mobile ligands mimic the in vivo system more faithfully, the use of immobilized ligands complements the former experimental methods by adding a myriad of potential patterning techniques that cannot be realized with mobile ligands. Specifically, in our system, the use of immobilized small ligands provided the spatial resolution required for distinguishing the effects of antigen density from its total amount. In line with ref 4, the present work demonstrates the tremendous value of submicrometer patterned surfaces to cell biology research. In particular, the combination of top-down technologies such as EBL with bottom up approaches such as SAMs offers unlimited possibilities for the research of receptor−ligand interactions and cell synapses.



ASSOCIATED CONTENT

S Supporting Information *

XPS analysis, additional figures, and methods section. This material is available free of charge via the Internet at http:// pubs.acs.org.



AUTHOR INFORMATION

Notes

The authors declare no competing financial interest.



REFERENCES

(1) Love, J. C.; et al. Self-Assembled Monolayers of Thiolates on Metals as a Form of Nanotechnology. Chem. Rev. 2005, 105 (4), 1103−1170. (2) Houseman, B.; Mrksich, M. The microenvironment of immobilized Arg-Gly-Asp peptides is an important determinant of cell adhesion. Biomaterials 2001, 22 (9), 943−955. (3) Senaratne, W.; et al. Functionalized surface arrays for spatial targeting of immune cell signaling. J. Am. Chem. Soc. 2006, 128 (17), 5594−5595. (4) Arnold, M.; et al. Activation of Integrin Function by Nanopatterned Adhesive Interfaces. ChemPhysChem 2004, 5 (3), 383−388. (5) Smith-Garvin, J. E.; Koretzky, G. A.; Jordan, M. S. T Cell Activation. Annu. Rev. Immunol. 2009, 27 (1), 591−619. (6) Aivazian, D.; Stern, L. J. Phosphorylation of T cell receptor [zeta] is regulated by a lipid dependent folding transition. Nat. Struct. Mol. Biol. 2000, 7 (11), 1023−1026. (7) Minguet, S.; Schamel, W. W. A. A permissive geometry model for TCR-CD3 activation. Trends Biochem. Sci. 2008, 33 (2), 51−57. (8) Davis, S. J.; van der Merwe, P. A. The kinetic-segregation model: TCR triggering and beyond. Nat. Immunol. 2006, 7 (8), 803−809. (9) Gil, D.; et al. Recruitment of Nck by CD3ε Reveals a LigandInduced Conformational Change Essential for T Cell Receptor Signaling and Synapse Formation. Cell 2002, 109 (7), 901−912.

4995

dx.doi.org/10.1021/nl302619p | Nano Lett. 2012, 12, 4992−4996

Nano Letters

Letter

(10) Artzy-Schnirman, A.; et al. Electrically Controlled Molecular Recognition Harnessed to Activate a Cellular Response. Nano Lett. 2011, 11 (11), 4997−5001. (11) Eshhar, Z.; et al. Specific activation and targeting of cytotoxic lymphocytes through chimeric single chains consisting of antibodybinding domains and the gamma or zeta subunits of the immunoglobulin and T-cell receptors. Proc. Natl. Acad. Sci. U.S.A. 1993, 90 (2), 720. (12) Roberts, C.; et al. Using Mixed Self-Assembled Monolayers Presenting RGD and (EG)3OH Groups To Characterize Long-Term Attachment of Bovine Capillary Endothelial Cells to Surfaces. J. Am. Chem. Soc. 1998, 120 (26), 6548−6555. (13) Bamdad, C. The Use of Variable Density Self-Assembled Monolayers to Probe the Structure of a Target Molecule. Biophys. J. 1998, 75 (4), 1989−1996. (14) Sigal, G. B.; et al. A Self-Assembled Monolayer for the Binding and Study of Histidine-Tagged Proteins by Surface Plasmon Resonance. Anal. Chem. 1996, 68 (3), 490−497. (15) Wen-Hsi, C.; Jui-Che, L. Surface characterization and platelet adhesion studies for the mixed self-assembled monolayers with amine and carboxylic acid terminated functionalities. J. Biomed. Mater. Res., Part A 2007, 82A (4), 820−830. (16) Lee, I.; Wool, R. P. Controlling amine receptor group density on aluminum oxide surfaces by mixed silane self assembly. Thin Solid Films 2000, 379 (1), 94−100. (17) Rusconi, S.; Kohler, G. Transmission and expression of a specific pair of rearranged immunoglobulin μ and κ genes in a transgenic mouse line. Nature 1985, 314 (6009), 330−334. (18) Jena, B.; Dotti, G.; Cooper, L. J. N. Redirecting T-cell specificity by introducing a tumor-specific chimeric antigen receptor. Blood 2010, 116 (7), 1035−1044. (19) Kochenderfer, J. N.; et al. Eradication of B-lineage cells and regression of lymphoma in a patient treated with autologous T cells genetically engineered to recognize CD19. Blood 2010, 116 (20), 4099−4102. (20) Doh, J.; Irvine, D. J. Immunological synapse arrays: patterned protein surfaces that modulate immunological synapse structure formation in T cells. Proc. Natl. Acad. Sci. 2006, 103 (15), 5700−5705. (21) Shen, K.; et al. Micropatterning of costimulatory ligands enhances CD4+ T cell function. Proc. Natl. Acad. Sci. 2008, 105 (22), 7791−7796. (22) Mossman, K. D.; et al. Altered TCR Signaling from Geometrically Repatterned Immunological Synapses. Science 2005, 310 (5751), 1191−1193. (23) Shen, K.; et al. Self-Aligned Supported Lipid Bilayers for Patterning the Cell-Substrate Interface. J. Am. Chem. Soc. 2009, 131 (37), 13204−13205.

4996

dx.doi.org/10.1021/nl302619p | Nano Lett. 2012, 12, 4992−4996