Figure 2. Screening a population of cells for those producing GFP. (Click to enlarge)

The cells are firmly anchored in the gel matrix. Supplementary Figure S2 shows that neither the density nor the morphology of the immobilized cells appreciably changed during bathing the gel in a saline solution up to 3 days. This allows the cells in the array to be interrogated or synchronized (22) by merely adding an effector to the surrounding solution or changing the solution. It also enables long-lasting studies, including real-time or time-lapse observations of many individual cells in large populations, such as studies on intracellular dynamics (23). Importantly, cells become immobilized irrespective of their adhesive properties.

Incubation of the arrays in a complete growth medium resulted in cell growth and division (Figure 3 and Supplementary Figures S3–S5). Both the nonadherent cells (such as lymphoblasts; Supplementary Figure S3) and adherent cells (typical 2-D culture cells; Supplementary Figures S4 and S5) produced spheroid microcolonies characteristic of a 3-D culture. Unlike the on-top 3-D cultures (16, 17), here cells are entirely embedded in the agarose matrix, which provides for secure immobilization of microcolonies during several days of culture. The PAA bed is chemically inert and presumably does not contact directly with the cells suspended in the agarose matrix. Yet, the PAA bed may mechanically influence the nearby cells, and this can be used to control the cell physiology through varying the PAA stiffness by changing the acrylamide and bisacrylamide concentrations (10).

Figure 3. Time course of HEK-293 cell growth in merged gels containing complete growth medium. (Click to enlarge)

Instead of agarose, other currently used 3-D matrices, both natural and synthetic (3, 4, 24-28), can be used to immobilize cells, or agarose can be supplemented with any desired component of ECM to better suit the requirements of a particular cell type (11, 29-32). For example, the presence of bovine skin collagen (consisting of type I and type III collagens) in the agarose matrix promoted the growth of some adherent cells and influenced the colony morphology at the later stages (compare Supplementary Figure S5, A and B). Importantly, unlike the PAA-based on-top cultures (16), no covalent linkage of the collagen molecules to the PAA gel is needed here, as those molecules become mechanically entrapped within the agarose matrix (33).

Observation of growing GFP-producing HEK-293 cells with a confocal microscope (Figure 3B) revealed no lag in the colony formation, suggesting that the low temperature used for agarose gelling did not appreciably harm the cells. The doubling time was ╛1 day, the same as in the conventional 2-D culture, and after 7 days, a colony contained up to 200 cells. Colonies of this size were still retained by the agarose matrix. This allowed the contents of colonies to stay separate and allowed the cells from individual colonies to be accurately withdrawn for further use, in particular for isolating cell clones.

Importantly, our approach can also be used to generate stable cell lines by cloning. We started with a population of the HEK-293 cells transfected by a GFP-encoding plasmid that could not replicate inside the cells and contained no sequences promoting its integration into the chromosomes. Furthermore, we did not use antibiotic or any other form of chemical selection that could affect the cell physiology or result in cell line instability (34, 35) nor used a cell sorter to enrich the population by GFP-producing cells.

Approximately 30,000 transfected cells were arrayed at a density of ╛100 cells/mm² (Figure 4A) in two PAA/agarose gels and allowed to grow for 7 days (Figure 4B). Then, fluorescing microcolonies that appeared homogenous (like those pinpointed with arrows in Figure 4B) were extracted from the agarose with a micromanipulator-handled glass capillary. As a rule, a colony was pulled out as a whole, with no cells left behind. The extracted colonies were individually transferred into wells of a standard 96-well plastic plate and incubated there for 7–10 days. Upon transfer, a colony flattened on the well bottom and continued to grow on the plastic surface as a 2-D culture. In the first experiment, whose results are presented in Figure 4, we picked 43 green colonies, 18 of which propagated (42% survival rate). [The survival rate was increased to 73% (66 of 91) and 97% (62 of 64) in the second and third experiments, respectively, owing to a more gentle manipulation of colonies at extraction.] Of those, we chose six variants by visual inspection, looking for fluorescence intensity, homogeneity, and growth rate; the latter was estimated as the time required for a colony to cover ╛20% of the well bottom.