Reverse transfection on cell arrays is a high-throughput method for the parallel transfection of mammalian cells for use in high-content screening light microscopy. Here, we present novel 9216-microwell cell arrays which combine the advantages of multiwell plates (physically separated samples) and cell microarrays (high sample density and long-term storage).

High-throughput transfection and highly correlative methods are the basis for cell-based systems biology. Reverse transfection on cell arrays (1) is a high-throughput method for the parallel transfection of mammalian cells. GFP-tagged cDNAs or small interfering siRNAs (2,3,4,5,6) are printed together with a gelatin solution at defined locations on glass slides. The transfection reagent can be printed either together with the nucleic acid/gelatin mix, or can be added to the printed array prior to cell seeding. We use the variant of printing the nucleic acid/gelatin together with the transfection reagent which, after drying, results in a “ready-to-transfect” platform. Tissue culture cells are plated on the dried slides resulting in clusters of live cells with a specific gene perturbation in a lawn of non-transfected cells, a method termed ‘reverse transfection.’ The miniaturized format of the array allows high-speed data acquisition (e.g., time-lapse data acquisition) of many samples in parallel (7). Importantly, the dried and ready-to-transfect replicates can be stored for >15 months (8).

Important disadvantages of the standard array-based reverse transfection technique are the risk for cross-contamination and the lack of reference markers when running fully automated screening pipelines. In addition, the total number of spots (siRNAs) per slide should be maximal for high-speed and parallel data acquisition. We introduce herein next-generation microwell cell arrays with physically separated cavities on the glass slide using a titanium coating. The cell array dimensions, including the connected polystyrene frame for cell solution handling, are according to SBS plate standards (Figure 1). The glass substrate (82.2 mm × 124.6 mm) consists of borosilicate glass (D263; Berliner Glas KGaA, Berlin, Germany), with a thickness of 500 µm supporting the standard optical methodologies. The optimized array layout resulted in 9216 spots per glass slide. These spots previously were defined by a standard photolithographic process. Photoresist (AZ-Lack; Allresist GmbH, Strausberg, Germany) was spread on the substrate and exposed to UV light through an array-patterned mask (MaskAligner; Suss Microtech Lithography GmbH, Garching, Germany). Exposed photoresist was removed by an organic solvent, whereas unexposed photoresist remained on separated cyclic spot areas with a diameter of 400 µm. After coating the array with a titanium layer of 200 nm by means of ultra-high vacuum evaporation (Classic 590; Pfeiffer Vacuum AG, Asslar, Germany), the photoresist fields were removed by a lift-off process. Thus, cavities suitable for reagent addition were generated with a free glass bottom and 200 µm–deep titanium layer with distances of Δx = 1125 µm and Δy = 750 µm, respectively. Three sets of reference markers at the borders of the glass slide (Figure 1, see arrow for one set of the markers) allow for precise location of the array coordinates during different processing steps, ranging from automated probe preparation to automated data acquisition. For highly automated screening pipelines, such a dense arrangement of spots will be required (Figure 2A). Other advantages of cell arrays are their unique high capacity for parallel assays, low cost (which is comparable to that of standardized glass slide well plates), and the flexibility in automated processes: depending on the required resolution, substrate thicknesses ranging from 700 µm [10× objective lens; numerical aperture (N.A.) = 0.4] to at least 200 µm (63× objective lens; N.A. = 1.4) are possible. Compared with commercially available plate formats with 1536 wells maximum, the gain in throughput by the microwell cell array presented here is a factor of six.

Figure 1. 9216-microwell cell array for high-content screening microscopy. (Click to enlarge)

Figure 2. Cavities on top of the glass slide and phenotypic data. (Click to enlarge)

To demonstrate the feasibility of the microwell cell array, we compared the transfection efficiency of solid phase reverse transfection by phenotypic penetrance following standardized protocols as described previously (8,9). siRNA oligonucleotides primers INCENP (inner centromer protein) and PLK 1 (polo-like kinase 1)—which target established mitotic events—and non-silencing control siRNA (Ambion, Austin, TX, USA) were used for preparing the source transfection solution. The solute compounds were spotted into the predefined array cavities in a highly parallel process using robotic-aided contact printing technology. Thus, the distribution of different compounds was managed in a pure and highly reproducible way within two hours for a single array. After drying the transferred compounds and fixing the frame, a ready-to-transfect microwell cell array was obtained, carrying 9216 samples displayed on top of the glass substrate of each of the cavities. HeLa cells were manually seeded upon the entire array in full growth medium using a pipette. At 24 h following plating, nuclei were stained with Hoechst stain (Cat. no. B2261; Sigma-Aldrich, Steinheim am Albach, Germany).

The data was then acquired with an automated scanning microscope (Olympus IX81 Scan^R;Olympus Biosystems, Munich, Germany) equipped with a 10×/0.4 N.A. air objective lens (Cat. no. UPSLAPO; Olympus) and a GFP bandpass filter (41017 Endow GFP Bandpass Emission; Chroma Technology, Fuerstenfeldbruck, Germany), with an overall imaging time of ~10 h for the whole microwell cell array. Figure 2, B–G, depicts representative phenotypes corresponding to the target-specific primers INCENP and PLK, as well as control siRNA. These siRNA-specific results are consistent with previous publications (7): INCENP knockdown resulted in multilobing of the cell nuclei due to segregation defects (Figure 2E) and PLK 1 knockdown resulted in the mitotic phenotype of prometaphase arrest (Figure 2G). In contrast, the non-silencing control siRNA had no impact on the shape of the nuclei (Figure 2C). Importantly, the appropriate phenotypes observed in neighboring spots in regions with different siRNAs indicate a lack of cross contamination.

In summary, we successfully adapted standardized protocols for solid-phase reverse transfection to a completely new microwell cell array, enabling up to 9216 different siRNAs to be transfected in parallel. The transfection solution was reliably printed inside the cavities using customized spotting techniques and, despite the high spot density of the array, no cross-contamination in terms of cell growth outside the cavities was observed. Thus, the cell array presented here showed its flexibility and applicability in advanced automated high-content screening pipelines. Moreover, the thickness of the glass slide is variable (200–700 µm), making it amenable to more advanced experimental setups (e.g., using objective lenses with higher numerical apertures for signal detection). The lack of cross-contamination at this current spot density indicates the possibility of a further possible reduction in spot-to-spot distances, thus resulting in a higher number of spots per cell array, likely achievable by thickening the titanium coating to create deeper cavities (up to 500 nm).