Efficiency, error and yield in light-directed maskless synthesis of DNA microarrays

Abstract

Background: Light-directed in situ synthesis of DNA microarrays using computer-controlled projection from a digital micromirror device--maskless array synthesis (MAS)--has proved to be successful at both commercial and laboratory scales. The chemical synthetic cycle in MAS is quite similar to that of conventional solid-phase synthesis of oligonucleotides, but the complexity of microarrays and unique synthesis kinetics on the glass substrate require a careful tuning of parameters and unique modifications to the synthesis cycle to obtain optimal deprotection and phosphoramidite coupling. In addition, unintended deprotection due to scattering and diffraction introduce insertion errors that contribute significantly to the overall error rate.

Results: Stepwise phosphoramidite coupling yields have been greatly improved and are now comparable to those obtained in solid phase synthesis of oligonucleotides. Extended chemical exposure in the synthesis of complex, long oligonucleotide arrays result in lower--but still high--final average yields which approach 99%. The new synthesis chemistry includes elimination of the standard oxidation until the final step, and improved coupling and light deprotection. Coupling Insertions due to stray light are the limiting factor in sequence quality for oligonucleotide synthesis for gene assembly. Diffraction and local flare are by far the largest contributors to loss of optical contrast.

Conclusions: Maskless array synthesis is an efficient and versatile method for synthesizing high density arrays of long oligonucleotides for hybridization- and other molecular binding-based experiments. For applications requiring high sequence purity, such as gene assembly, diffraction and flare remain significant obstacles, but can be significantly reduced with straightforward experimental strategies.

Figure 1

Phosphoramidite synthesis cycle in maskless, light-directed synthesis of microarrays. The cycle is similar to that used in solid-phase synthesis of nucleic acids with some key differences: UV light from the I-line of mercury, in the presence of an organic base, is used to deprotect the 5'-OH; the array surface is dried with helium before photodeprotection; oxidation of the phosphites is not required in the cycle because they are not exposed to acid; the final chemical deprotection must not cleave the nucleic acids from the surface. The duration of each step in the synthesis cycle depends on experimental conditions and objectives, but typical values are given.

Figure 2

Scan image of 4-plex microarray used for coupling studies. Coupling yield based on up to four sets of coupling parameters (or in a separate experiment, different phosphoramidites) were determined on single microarrays. The labels A, C, G, and T/U can refer to the four amidites, but in this scan only NPPOC-dT was used with four different sets of coupling parameters. The numbers at the top and bottom label the length of the experimental oligonucleotide, which is synthesized after a thymidine 15mer linker. Each numerical label is associated with four features, labeled in white in the top left corner: Cy3-labeled n-mer (a), unlabeled n-mer (b), Cy3-labeled 1-mer (c), and unlabeled 1-mer (d). The normalized intensity I of each n-mer is calculated as [I(a)-I(b)]/[I(c)-I(d)]. The data from this array is given in Table 2, Array 4. The 0-mer features are the result of coupling from a phosphoramidite port on the synthesizer containing pure acetonitrile and are used to calibrate the capping efficiency. Spot intensities are uniformly high due to high coupling efficiency.

Figure 3

Coupling efficiency vs. time of NPPOC-dT phosphoramidite. Coupling efficiency starts at ~95% at the lowest possible transit time, increases to a maximum and then decreases slowly for long coupling times. Circles indicate a phosphoramidite concentration of 0.03 M. At this concentration and between 30 s and 150 s coupling times, no yield differences can be observed between 1-mers and 12-mers. Squares show coupling efficiency data for the phosphoramidite diluted to 75% of the original concentration. The curve serves as a guide to the eye (lognormal; μ = 52; σ = 6.9; scaled to 100% at the maximum at 80 sec.).

Figure 4

Single sequence exposure gradients with different oxidation protocols. Signal intensity is from a hybridized, Cy3-labeled complementary sequence. Black circles: single oxidation after last coupling. Red squares: an oxidation at the middle and one at the end. Green triangles: oxidation after every fourth coupling and after the final coupling. Blue downwards triangles: Oxidation after every base. Insert graph: no oxidation steps. Inset images: one of four replicate gradients on the array with a middle and terminal oxidation; exposure increases bottom to top and right to left.

Figure 5

Stepwise coupling efficiency of dT with or without a 30 s helium drying step following capping. All of the plots are from data from a single 4-plex coupling microarray. The results show that the helium significantly increases coupling yield and that the helium step only increases the coupling efficiency of the immediately preceding coupling step. Percent coupling efficiency is derived from two parameter exponential fits of the data. Error bars are the standard deviation of the replicates.

Figure 6

Schematic of the optical system of the maskless array synthesizer. A. High pressure mercury short-arc lamp. B. Dichroic mirror. C. Homogenizing light pipe. D. Shutter. E. Folding mirrors. F. Micromirror array. G. Offner relay primary mirror. H. Offner relay secondary mirror. I Reaction chamber illuminated by ON mirrors. J. Light dump from OFF mirrors.

Figure 7

Global flare is due to dust and imperfections of optical elements and leads to a homogenous background exposure of the synthesis surface. With a UV intensity at the center of the image plane (gray squares in insets), global flare is measured from ON mirrors at the periphery of the DMD (white border of insets), which do not image directly into the detector. Global flare increases almost linearly with the number of ON mirrors. The central square in the inserts represents the photo-detector and the outer bands represent the number of ON mirrors. For 25% ON mirrors the irradiance (W/cm²) value is about 0.04% of direct irradiance.

Figure 8

Mirror-edge scattering is a form of local flare that is imaged to the synthetic surface and leads to exposure in the interstices between the synthesis pixels for both ON and OFF mirror positions. A. Simulated edge scattering using Silvaco Optolith. B. Cy3-labeled, single dT microarray synthesized using long exposure only with OFF mirrors. C. Simulated exposure with ON mirrors. D. Cy3-labeled, single dT microarray synthesized using normal exposure with ON mirrors. The central dark regions are due to the mirror support post (part of the DMD mechanics), which both reduces mirror reflection in the center and introduces additional scattering.

Figure 9

Calculated aerial views of intensity on the synthetic surface from a 3 × 3 array of micromirrors. A. Only central mirror ON. B. Only central mirror OFF. C. All mirrors ON. D shows a logarithmic plot of the light intensity reaching a central horizontal line through A (dash), B (solid) and C (dot-dash). In all cases the edges of the illuminated synthesis pixels lack sharpness, and particularly for the all ON pattern, the gaps between pixels are receive substantial intensity, about 50% of maximum pixel intensity.

Figure 10

Dose calibration curve using direct fluorescent end-labeling with Cy3. A single base (dT) was coupled uniformly over the glass substrate and different areas (images at the bottom) were exposed to different doses of light before the final coupling with a Cy3 phosphoramidite. The time constant for photodeprotection τ equals approximately 2.5 J/cm².

Figure 11

Optimal photodeprotection dose for maximum correct sequence yield in microarray synthesis for gene assembly experiments. The curves are calculated from Eqs. 2-4 with the coupling efficiency (γ) set to a value of 100% since coupling is an independent factor and does influence the calculation of the optimal dose. The two functions, bright and dark exposure deprotection yields, which are multiplied to give the correct sequence yield curve, are shown with dotted lines for the 25mer with contrast ratio of 1/350 in the right panel.

Figure 12

A 3-D visualization of the MAS optical system with labels indicating the major components. B. Synthesis reaction chamber with arrows and colors indicating relative reagent velocity within the parallelogram-shaped volume defined by the gasket and across the microarray surface (central rectangle). The reagent flow pattern indicates that the cell geometry prevents stagnant zones which might hinder array uniformity. Relative to the illustration, when mounted in the synthesizer, the reaction chamber is rotated by 45° clockwise, with reagents entering at the bottom and exiting at the top (black spots). The scale of the chamber is defined by the rectangle, which, due to the 1:1 imaging system has the same dimensions as the DMD (14 × 10 mm).