The critical role of microscale cellular environments in regulating cell behavior has led to the development of a variety of biomicropatterning technologies. Most of these methods are adapted from silicon micromachining, polymer micromolding, stamping, or inkjet printing where the substrate on which the patterns are created is substantially dry and often hard. The canvas of life, however, is wet, soft, and alive. Is it possible to directly sketch microscale gene expression and suppression experiments onto the delicate surface of cells without compromising their viability? Here we show some of our initial biopainting results that support this view.

Introduction
Micropatterning of biological materials facilitates cell-material interactions [1, 2] and allows regulation of cell function [3]. Available techniques mainly utilize an elastomeric stamp [1, 4] or a solid pin [5] to transfer reagents such as cell adhesion molecules and gene constructs onto a dry surface upon contact. Subsequent overlaying of the micropatterned surface with cells generates an array of cellular phenotypes. Physical contact and the need for dry substrates limit these techniques for patterning onto delicate surfaces of living cells. Although laminar streams in microfluidic settings enable contact-free patterning of reagents over cells, the throughput is limited and reagent diffusion across laminar flow interfaces is inevitable [6]. Thus, techniques enabling non-contact reagent patterning on cells in fully aqueous media will greatly benefit various biological studies of cell-reagent and cell-material interactions.

Aqueous Two-phase Systems as a New Micropatterning Tool
To address this need, we have developed a new micropatterning strategy using a polymeric aqueous two-phase system (ATPS) with polyethylene glycol (PEG 8K, 4 % (w/w)) and Dextran (DEX 500K, 5 % (w/w)) as the phase forming polymers and cell culture media as the solvent (fig. 1a) [7]. Cells retain a high viability of >96 % in both aqueous phases. To form a pattern on cells, the DEX phase is loaded into a pipette tip and lowered to close proximity of the cell monolayer maintained in the PEG phase.

The denser DEX phase dispenses onto cells and simultaneous movement of the pipette tip allows formation of arbitrarily shaped patterns such as "UMICH" on cells (fig. 1b) [8]. Long-term pattern stability was demonstrated by including liposomal complexes of genetic materials in the patterning DEX phase and observation of resulting localized gene expression that exactly mimic the shapes of the patterned reagent [8]. An extremely low interfacial tension between the two aqueous phases (on the order of 0.01 mJ/m2) and interactions of cells with the patterned DEX phase are key to the stability of patterns.

Microarray Formation
We demonstrated the utility of this technology for microarray format high-throughput gene function studies. To form a microarray of genetic reagents on cells, first solutions of PEG and DEX were prepared in Optimem media, mixed together, and allowed to equilibrate to form an ATPS (fig. 1a). The two phases were then separated. For lipid-mediated gene delivery to cells, liposomal transfection complexes were prepared using lipofectamine 2000 and plasmid DNAs, mixed with the DEX phase and transferred to a 1536-well plate. Nanoliter dispensing pins were mounted on a pin tool fixture (V&P Scientific). Clean pins were dipped into the wells, filled with the solution containing transfection complexes, and slowly withdrawn from the solution. Then, pins were lowered into the close vicinity of the cell monolayer immersed in the PEG phase and were allowed to dispense the DEX phase containing genetic materials. After formation of array of droplets on cells, pins were slowly retracted and moved out of the culture dish. Cells were incubated for 6-8 hrs at this condition to allow uptake of genetic materials. Then the two-phase media was washed out and replaced with regular culture media. Cells were incubated for another 48 hrs to allow the expression of transfected genes and finally imaged using a fluorescent microscope. A similar strategy was used for lentiviral-mediated gene expression and knockdown studies except that (i) the ATPS was prepared in regular culture media and (ii) lentiviruses carrying cDNA or short hairpin RNA (shRNA) were used instead of lipid-plasmid complexes [8].

Gene Expression and Knockdown Microarrays
Figure 2a shows a 6×4 microarray of HEK293H cell clusters expressing eGFP, dsRed, or co-expressing both genes in a lawn of non-transfected cells. Each cluster of transfected cells was exposed to a 500 nl droplet containing only ~10 ng plasmid. We also showed patterning with volumes as small as 20 nl on cells to significantly reduce the amount of reagents required and increase the density of spots in the microarray. Our analysis showed that the highest level of protein expression, as measured by fluorescence intensity in the cell clusters, is obtained using a complex preparation with a 1:1 ratio of plasmid DNA to transfection reagent (µg/µl).