PCR-based setup for high-throughput cDNA library sequencing on the ABI 3700 automated DNA sequencer.

Many large-scale sequencing projects are now being conducted or considered that rely on the latest capillary electrophoresis-based automated sequencing system from Applied Biosystems, the ABI 3700 (PE Biosystems, Foster City, CA, USA). Because of the electrokinetic injection protocol and the physical properties of the capillaries and polymer used, reaction setups that have been optimized for the gel-based ABI 377 systems (PE Biosystems) do not necessarily perform to expectations in the capillary system. We found this to be especially so when sequencing PCR products. We describe a rapid, inexpensive and reliable method for preparing template and sequencing reactions for application to high-throughput sequencing on the ABI 3700 instrument. The overall goal of the procedure was to produce a sequence product with little contamination by salt or unincorporated fluorescent dyes while minimizing cost and effort. We had previously developed a robust protocol similar to the one described here, having a high (80%–90%) overall success rate on our ABI 377 systems. However, our initial attempts to use the same protocol on the ABI 3700 produced only short read lengths, primarily because the sequencing reactions were too successful, that is, they were producing more product than could be handled by electrokinetic injection. Also of concern was the observation that sequence quality dropped from the first run of a plate to the fourth, apparently because of degradation of the reaction in the formamide or water loading solution while the plate sat at room temperature on the instrument. A third concern was the possibility that fragments of the bacterial cells used as templates for the reaction might decrease the lifetime of the capillary array by blocking the capillaries. Several adaptations were necessary to reliably produce high-quality sequence reads on the ABI 3700. The starting materials for the sequencing reactions were arrayed, normalized cDNA libraries in 384-well plates. The libraries were constructed in the pSPORT6 vector (Life Technologies, Rockville, MD, USA) in which the cloning site is flanked by sequences corresponding to the standard M13, T3, T7 and SP6 sequencing primers. The average insert size in these normalized libraries was approximately 1000 bp, with a range from 100 bp to 4 kb. The individual clones were stored frozen in LB broth supplemented with 10% glycerol and 50 mg/L ampicillin. All manipulation steps on individual wells were carried out on a Biomek 2000 robotic workstation (Beckman Coulter, Fullerton, CA, USA) using MP20 and MP200 pipetting heads. Before amplification, plates were defrosted and replicated in a fresh 384-well growth plate (Marsh Biomedical Products, Rochester, NY, USA) containing 50 μL LB broth containing 50 mg/L ampicillin using a 384-pin high-density replicating tool (Beckman Coulter). We found the use of fresh cells, rather than glycerol stock cells, substantially increased the rate of successful amplification. The clones were grown overnight at 37°C while being shaken at 225 rpm, and a 10-μL aliquot transferred to a 384-well PCR plate. Plates were centrifuged at 2000× g for 20 min to pellet the cells. The medium was removed by inverting the plates and shaking them. Cells were resuspended in 50 μL 10 mM Tris (pH 8.0) by vortexing, and 1 μL transferred to a 384-well PCR plate containing 10 μL PCR mixture [0.1 mM each dNTP, 0.2 U Taq DNA polymerase (Amersham Pharmacia Biotech, Piscataway, NJ, USA)], 1.1 μL 10× Taq buffer and 0.5 μM each M13 forward and reverse sequencing primers. Plates were sujected to 30 cycles of amplification (94°C for 30 s, 60°C for 30 s and 72°C for 2 min), followed by centrifugation at 2000× g for 10 min to pellet remaining bacterial debris. To remove the remaining amplification primers from the sequencing mixture, a 2.5-μL aliquot of the reaction was transferred to a fresh 384-well sequencing reaction plate (PE Biosystems) containing 0.2 U exonuclease I (New England Biolabs, Beverly, MA, USA) in 7.5 μL water, mixed and incubated for 1 h at 37°C. The exonuclease was inactivated by incubation at 65°C for 20 min, and the amplification products were precipitated by the addition of 20 μL 100% ethanol and centrifugation at 1700× g for 30 min. Residual ethanol was removed by inverting the plate and centrifuging at 50× g for 2 min. Sequencing was initiated by resuspending the amplification products in 5 μL sequencing reaction mixture [0.5 μL BigDye Terminator Ready Reaction Mix (PE Biosystems), 1.5 μL halfB D sequencing reagent (Sigma, St. Louis, MO, USA) and 3 μL 1 pM SP6 sequencing primer] and subjected to 25 cycles of 96°C for 30 s, 50°C for 1 min and 60°C for 4 min. Sequencing reactions were precipitated by the addition of 20 μL 70% isopropanol with mixing and centrifugation at 1700× g for 30 min at room temperature. Residual isopropanol was removed as described above. Pellets were washed with 20 μL 70% ethanol, with another cycle of centrifugation and removal of residual ethanol. Plates were air dried for 5–10 min and stored in a desiccator at -20°C. Immediately before being placed in the loading chamber of the sequencer, 30 μL ultrapure formamide (PE Biosystems) were added, and the plates were spun at 670× g for 2 min at room temperature. We did not find it necessary to heat denature the samples before loading them onto the sequencer. Parameters for running the instrument with POP-6 polymer are shown in Table 1. We found that replacing the growth media with fresh buffer significantly increased the success of the clone amplification step for our normalized libraries. It is likely that this is an effect of the host strain used for normalization of the library because a separate, commercial, non-normalized bovine lymph node library (Stratagene, La Jolla, CA, USA) did not show the same effect and this step was unnecessary. However, the media replacement step is inexpensive and does not significantly increase the time required for the process. The procedure described here has an amplification success rate of approximately 85%–95%, which is not significantly different than the success rate for template preparation by automated miniBenchmarks