The POSaM platform. (a) Overview. The complete inkjet printing system is enclosed in an air-tight acrylic cover, 61 × 91 × 122 cm. (b) View from above showing the array holder. One slide is shown secured by the vacuum check with room for 26 additional slides. (c) Front view showing the print/wash head. Five PTFE wash lines deliver acetonitrile, oxidizer and deprotecting acid in bulk. Six vials supply tetrazole and phosphoramidites to the inkjet print head. (d) Lower-front view of the inkjet print head showing droplets passing through the QC laser beam. The presence of a droplet produces forward-scattered light, visible as bright red flashes (arrowed).

Diagram of the POSaM. The system uses two PCI-interface input/output boards (MIO-16E and DIO-32HS) and one ethernet-interface servo controller (6K4). Future software revisions will support a second print head, to provide more reagent channels, and the X-axis linear position encoder to increase printing speed.

Design of quality-control droplet-detection subsystem. Before each cycle of synthesis the inkjet print head is moved over the light beam of a red diode such that the stream of droplets ejected from each pump intersects the beam at right angles. (a) If a nozzle is defective, there is no droplet ejected into the light beam, light is not scattered and the narrow beam is absorbed by a bar placed before the lens of the photomultiplier tube. (b) If the nozzle is functioning, a droplet intersects the light beam and scatters the light, which is then focused on the photomultiplier tube (PMT) to signal that the nozzle is functioning correctly. The state of the droplet is held by a D-type flip-flop. The circuit is reset just before droplet ejection and the state is read immediately after. If nozzle failures are detected, the software reschedules the motion and firing of the printing head so that the most efficient printing path is taken.

Photomicrograph of virtual reaction wells on silanized glass. Phosphoramidites and tetrazole dissolved in 1:1 3MP:MGN were printed onto a standard epoxysilane-modified slide surface. The slide was carefully removed from the POSaM instrument and a photomicrograph made using 50 × magnification of a light stereomicroscope with a cooled-CCD digital camera (1,392 × 1,040 resolution). ImageJ was used to draw the scale line on the photomicrograph.

Photomicrographs of virtual well successes and failures. Slides were printed and photographed as stated in Figure 4 with the following modifications. (a) Virtual reaction wells are each formed by printing a droplet of phosphoramidite solution followed by a droplet of tetrazole solution using the standard 1:1 3MP:MGN. (b) Tetrazole droplets were printed before phosphoramidite droplets. Discrete virtual reaction wells fail to form. (c) Wells formed by printing two droplets of phosphoramidite solution. (d) Two droplets of tetrazole solution printed with no phosphoramidite. Virtual wells fail to form. (e) Phosphoramidite and tetrazole dissolved in 1:3 3MP:MGN and phosphoramidite printed first. Virtual wells form well. (f) Two droplets of tetrazole in 1:3 3MP:MGN solvent. Although not perfectly round, virtual wells do form.

Chemical labeling of an oligonucleotide array. Four sets of 25-base oligonucleotides were synthesized in triplicate on standard epoxysilane-modified glass slides using the 1:1 3MP:MGN solvent system, printing phosphoramidites before tetrazole. Each oligonucleotide contained from 1 to 10 guanosine residues. Alexa-594 dye was conjugated to the guanine bases using the instructions included in the ULYSIS kit (Molecular Probes), and the slide was scanned on the ScanArray using 10 μm resolution at standard laser and PMT settings. Fluorescence intensity was determined using the Dapple spot-finding program. (a) Laser scan of the array. The spots surrounding the experimental Alexa-labeled spots are standard gridding spots that are prehybridized to a control Cy5-labeled oligonucleotide, and are included on all arrays to facilitate grid placement and as a check on virtual well formation. (b) Fluorescence intensity from Dapple output graphed against the number of dGMP residues in the molecule. A good linear fit is observed (r² = 0.99) and shows fluorescence intensity to be proportional to guanine content between four and 10 bases.

Sensitivity of mismatch detection. Slides were printed and fluorescence measured using the standard methods described in Materials and methods. (a) Fluorescence from reporters containing from zero to four mismatches to the target 12 mer (5'-Cy5-GCG TTG GCA CTG). Mismatched bases are in bold. (b) Fluorescence from reporters containing from zero to six mismatches to the target 20 mer (5'-Bodipy-GAC CTC CCG GAC ACG CAC CT). A single mismatch reduces the binding of the 12 mers by 64% and of the 20 mers by 25%. Mismatched positions are in bold.

Effects of mismatch type and position on inkjet oligonucleotide hybridization. Slides were printed using standard conditions described in Materials and methods. Reporter sequences representing all possible single-base mismatches in a 12 mer duplex were used. (a) The substitution of a non-guanine with a guanine (G) within the probe sequence was slightly, but significantly, less destabilizing than the other types of substitutions. (b) The destabilizing effect of the mismatch was less at the very ends of the probes than in the central positions. Position 1 is closest to the substrate; position 12 is furthest away.

Sensitivity to mismatches as demonstrated by two-color hybridization. An array was synthesized containing reporters complementary to targets 5'-Cy3-GAC CTC CCG CTC ACG CAC CT (green) and 5'-Bodipy-GAC CTC CCG GAC ACG CAC CT (red), using standard conditions described in Materials and methods. The bases in bold are different in the two reporter:target molecules. The slide was hybridized with an equimolar mixture of the two labeled target oligonucleotides and the fluorescence intensity determined on the ScanArray 5000 using the Dapple spot-finding program. The green/red ratios of the green- and red-complementary features were 3.12 ± 0.57 (N = 230) and 0.25 ± 0.05 (N = 660), respectively. Graphing Cy3 intensity vs Bodipy intensity shows excellent separation between signals and the insert shows that the effects of two base changes are clearly distinguishable by eye.

POSaM oligoarray re-use. Arrays of our 20 mer gridding sequences were synthesized as described in Materials and methods. The arrays were then hybridized to Cy3- or Cy5-labeled complementary target molecules and the fluorescence intensity determined as described above. Intensities were recorded and the slides were stripped of target DNA by incubation in 20 mM NaOH for 2 min at room temperature. The fluorescence intensity of the stripped slides was measured and the slides were rehybridized using the same targets as before. These hybridizations, intensity measurements, stripping, intensity measurements, and rehybridizations were repeated at least 10 times. Stripping reduced fluorescence by an average of 99%. Signal intensity and specificity appears unchanged during the first four cycles of use. The sequences used are identical to those in Figure 8: Cy3-labeled targets are perfectly complementary; Cy5 differ by two adjacent mismatches.

Detection of yeast deletion-strain barcodes from genomic DNA. Barcodes from 94 strains were PCR amplified and labeled with Cy3 and with Cy5, combined, and hybridized to an inkjet array. Approximately 10-fold additional genomic DNA from YPL110C, YPL111W and YPL112C was spiked into the Cy3 PCR, resulting in a 10-fold increase in Cy3 fluorescence from the corresponding reporters (gray bars). The array was stripped and the experiment was repeated, spiking the three strains into the Cy5 PCR. This resulted in a four- to fivefold increase in Cy5 fluorescence from the corresponding reporters (white bars). The laser-scan images in the inset show the increased Cy3 (green) signal from the YPL111W spike reporter in the first hybridization and increased Cy5 (red) signal from the YPL111 W spike reporter in the second hybridization. Signal is detected from the mismatch reporter as well, but at a greatly attenuated level. The slides used as substrate in this experiment have been modified using our own method as described in Materials and methods.