Prior to performing this analysis of chip-eluted oligonucleotides, modifications of the maskless array synthesizer were performed to allow for high-quality, high-density synthesis of long oligonucleotides onto glass slides. The enhancements utilized existing technology such as an image-locking system to allow for the improved, accurate synthesis of long oligonucleotides up to 90 bases in length, as shown in imaging and fluorescence studies [(13
); C. Kim, M. Li, M. Rodesch, A. Lowe, K. Richmond, N. Venkaratamaiah and F. Cerrina, submitted for publication]. This technique allows for the synthesis of not only longer oligonucleotides, but also more accurate oligonucleotides (by reducing inaccurate deprotection which causes subsequent base addition and/or deletion), and the ability to create more sequences per chip (reducing the need for ‘border’ areas to limit crossover deprotection of adjoining regions from drift).
One significant finding was the dramatic effect noted by the alteration of a single synthesis parameter on the quantity of full-length oligonucleotides produced. It is evident that the ‘standard’ method for DNA synthesis, although adequate for microarray production and use, may not be the best manner in which to synthesize large quantities of full-length oligonucleotides. Analysis of the synthesis products of the homopolymer T10
(Figure , lane 2) revealed that only a minority of oligonucleotides were full length (<35% of population). This was interesting since the coupling efficiency for NPPOC-T had been reported previously (98%) and for a T10
, therefore, 76% of products were expected to be full length. Such differences between the theoretical and experimental full-length oligonucleotides production could be accounted for by a myriad of reasons such as inefficiencies in deprotection (4
), coupling, or cleavage from the chip surface. Prior to this work, no analysis of chip eluates had been done on chips generated with NPPOC chemistry. These results led us to optimize the DNA synthesis parameters (coupling, exposure time, protocol steps) used in this study.
The difference in profiles between mixed base and poly(T) oligonucleotides (Figure a) was significant and intriguing. Although the efficiencies of base addition have been reported previously (11
) and shown to vary for the different NPPOC phosphoramidites (A-96%, G-97%, T-98% and C-99%), it is unlikely that these differences alone account for the decrease in M12
full-length products. Further optimization studies were performed to increase the amount of full-length product and it was noted that a 100-fold increase occurred not by changing the basic chemistry or reagents used, but through changes in the synthesis protocol itself. The combination of increased coupling time (from 20 to 60 s), increased exposure time(from 50 to 150 s), the addition of Ar drying steps, and an image-locking system was used to produce more full-length product even for longer oligonucleotides such as 40mers (Figure b).
Interestingly, chips that were not optimized for DNA synthesis yielded strong hybridization profiles (data not shown). Despite the lack of full-length oligonucleotides seen in gel profiles, hybridization appears sensitive enough to detect these species, present in only small percentages, and, therefore, hybridization results do not accurately determine overall oligonucleotide quality; they are merely indicative of trends in purity. In high-density arrays there are also additional factors [e.g. steric hindrance (21
)] which can affect the interpretation of hybridization results and make these results unreliable for any predicative use for chip oligonucleotide quality.
Initial experiments with oligonucleotides (41mers) from ‘optimized’ syntheses on base-labile slides were successful in assembling to form a larger DNA fragment (61mers). This is the first reported use of the base-labile linkage for release of DNA from chips and the use of eluted chip DNA in a biological application. Assemblies using amplified and digested oligonucleotides (eight 70mers digested to yield eight 40mers) were also successful and this method could be applied to create large populations of oligonucleotides for other biological purposes. The ‘AACED’ amplifying method has the advantage of not only increasing the number of oligonucleotides for use, but it also allows for the enrichment of the full-length product over truncate species. Additionally this method could allow for amplification of sub-populations of DNA from the chip surface, making each chip useful for a variety of oligonucleotides syntheses. Indeed, with the (overly) optimistic assumption of obtaining ~10 pmol/cm2
of oligonucleotides from a chip, this quantity would correspond to ~12 attomol/pixel and thus without amplification, most biological applications would not be feasible. If our AACED was extended to its full potential, a single chip with 786
432 unique 40mer oligonucleotides could potentially be used to assemble >15 Mb of DNA.
Miniaturization of hardware provides a means to reduce the quantities of reagents and therefore the costs of oligonucleotide synthesis. Technological applications using oligonucleotides, however, are becoming very diverse and hence the variety of sequences needed is escalating. In keeping with this trend, we have utilized a Biological Exposure Synthesis System (BESS) to produce chips with small amounts of synthesized oligonucleotides for subsequent elution and utilization in biological assays. This system can allow for hundreds of thousands of discrete oligonucleotides species to be made inexpensively and quickly for utilization in a myriad of biological assays, including that of gene assembly.