Abstract
The design of a fully automated high-throughput system for the purification of sequencing templates is described. Hardware, software, and chemistries have been optimized to suit the needs of high-throughput laboratories involved in genome sequencing projects. Using this system, up to 5760 samples (60 times 96-well plates) can be purified in less than 16 hours during a single unattended run. The system can also be configured to perform sequencing reaction setup for all 60 plates following template purification, extending total run times to < 24 hours. Final sequencing reactions are prepared in 384-well microplates.
INTRODUCTION
During recent years most of the public interest was focused on the Human Genome Project.1,2,1 However, a number of other model organisms and genomes are currently under investigation.4,5,6 Future efforts will be directed towards animal and human pathogens, in an effort to improve prevention and treatment of infections, and food crops in order to combat diet deficiencies, such as vitamin A deficiency, and reduce crop failures. With the number of samples predicted to increase considerably, there is a strong need for fast, fully-automated and cost efficient processing of front end procedures such as template purification and sequencing reaction setup, matching the throughput capacities of modern sequencing technologies.
A number of commercially available stand-alone workstations as well as integrated systems and ‘home-made’ automation solutions have been developed for nucleic acid preparation during the last years. Sample processing using high-throughput instruments usually does not include steps subsequent to nucleic acid purification.
In this report we describe the development of an high-throughput system for automation of front-end tasks in genome projects, including template purification and sequencing reaction setup. Our goal was to design a walk-away system that is capable of providing sufficient sample throughput to continuously feed multiple 96-lane sequencers. Specifications for quality of the obtained DNA were set to meet the more stringent requirements of the latest capillary electrophoresis DNA sequencing machines. Specifications for DNA yield allowed for running at least two sequencing reactions (200–500 ng DNA per sequencing reaction), including enough DNA for repeat sequencing if required and archiving.
NUCLEIC ACID PURIFICATION TECHNOLOGY
Components and layout of the automated system were largely dependant on the chosen nucleic acid purification technology. In choosing a technology for our project, cost aspects, automation friendliness, and chemistry characteristics such as DNA binding capacity and recovery rates were considered.
Magnetic bead-based nucleic acid purification chemistries are a cost-effective alternative to commonly used anion-exchange chromatography or silica-gel-membrane technologies, giving reduced running costs for high-throughput projects. Moreover, such chemistries are better suited to full automation because they require no centrifugation or application of pressure or vacuum to move solutions through a chromatography bed.
The QIAGEN® MagAttract™ 96 Miniprep System uses silica-coated magnetic beads to purify plasmid DNA directly from crude bacterial lysates. Novel lysis chemistries produce a clear lysate, eliminating the need for centrifugation or filtration steps. The purification procedure relies on the high affinity of plasmid DNA for silica particles under specific denaturing conditions. The beads consist of a ferromagnetic core that is coated with silica using an innovative manufacturing process. Each bead has a large surface area (Fig. 1), providing a high nucleic acid binding capacity (3 μg per 20 μl bead suspension). The beads show a strong induced magnetic moment when placed near a magnet. Magnets for separation in 96-well formats consist of a base with 24 vertical magnetic rods, onto which a microplate is placed. Each magnetic rod fits in the space between four wells. Beads and associated plasmid DNA are rapidly concentrated at the side of the well nearest to the magnetic rod, even in viscous bacterial lysates, ensuring fast and complete separation (Fig. 2). In the absence of a magnetic field, the beads become nonmagnetic.

MagAttract magnetic beads before (A) and after (B) separation.
This reversible magnetization avoids clumping of beads, allowing fast and efficient resuspension.
This novel chemistry was thoroughly validated using a stand-alone robotic workstation and gave mean read-lengths of 630 bp using an ABI PRISM® 3700 DNA analyzer (Applied Biosystems, Foster City, CA) with POP™-5 gel matrix, at a 99% level of accuracy (Fig. 3).

Phred 20 analysis of 1722 shotgun clone sequences from plasmid DNA isolated for a bacterial genome project. Clones were prepared using the MagAttract 96 Miniprep System on a stand-alone workstation, sequenced using BigDye™ Terminator chemistry, and analyzed using an ABI PRISM® 3700 DNA Analyzer with POP™-5 gel matrix. The analysis shows the percentage of sequence reads, which were calculated using sequential windows of 10 bases.
AUTOMATION DESIGN
The high-throughput system designed to fully automate purification of plasmid DNA using MagAttract 96 Miniprep technology, consisted of the following components (Fig. 4):

Overview of the high-throughput nucleic acid purification system.
A Zymark SciClone™ 96/384 ALH workstation, equipped with a 96-channel pipettor head and two 8-channel dispenser heads. The 96-channel pipettor head was used for plate-to-plate transfer steps and removal of supernatants using standard 200 μl pipetting tips. Tips were thoroughly washed between steps using distilled water. High-precision motion control allowed pipetting in 384-well plate formats. The 8-channel dispenser heads use precision syringe pumps with an accuracy of < 3% (CV) when dispensing volumes > 20μl and are used for distribution of Buffer DW4 and Buffer EB (see the protocol outline below).
Two Twister™ II robotic arms, capable of moving blocks and microplates between individual system components. One Twister II was used for moving labware only, whereas the other was equipped with storage pods and stackers for interim storage of processed microplates. Twister II robotic arms have six degrees of freedom and can manipulate microplates and other labware up to 0.5 kg in weight.
Two dispenser/shaker stations for reagent addition, resuspension of bacterial cell pellets, and separation of magnetic beads (Fig. 5). The dispenser/shaker stations were developed to streamline workflow and save laboratory space. Performing dispensing and shaking steps on a single unit reduces the number of robotic arm interactions, which reduces overall processing time. These stations perform shaking at up to 1100 rpm with an eccentricity of 3 mm and are capable of dispensing reagents with accuracy of < 3% (CV) for volumes >20μl.
The newly developed dispenser/shaker station.
A Heraeus Cytomat® 6001 (Kendro, Newtown, CT) incubator for storage of deep-well blocks and microplates. Storage temperatures of 4°C-RT at up to 95% relative humidity reduce evaporation of samples. The incubator provides storage capacity for more than 60 deep-well blocks or labware of similar dimensions. Storage of labware containing bacterial cell pellets, 384-well plates containing final DNA eluates, or sequencing reactions is possible.
A dedicated shaker station for sample incubation and binding of template DNA to silica-coated magnetic beads.
A digital control unit to ensure timely and coordinated functioning of all systems under control of CLARA 2001 software (Zymark Corporation, Hopkinton, MA). CLARA 2001 system software is a flexible software package for automated system integration. The software has been specifically designed to support applications in areas such as genomics, drug discovery, pharmaceutical development and quality control.
CLARA contains a Method Editor interface that provides drag-and-drop method development. Both sequenced and scheduled methods can be visualized in CLARA Gantt Charts (Fig. 6). The Run-Time Execution Manager allows to monitor, control, and define error handling during a method run. CLARA 2001 can map data output, export multiple concurrent data threads in real-time; export to ASCII, Excel and/or ODBC compliant data base files; and export data to a remote site via an intranet or website. The control box is equipped with a built in power sup ply for the RS-232 control of valves, pumps, switches etc. It includes: sixteen 24V or V inputs via NPN or PNP; sixteen 24V 500mA outlets; four 24V, 3A outputs; four 110–240 VAC, 10A outputs.
Overview of the high-throughput nucleic acid purification system.
An industrial aluminum profile worktable to accommodate all system components. The table included a leg set and a shelved bench top for the positioning of the system. Accessories, buffer carboys, or circulators can be placed on the floor underneath the table; the table is pre-wired with two separate 20 A electrical circuit breakers (120 V), including receptacles for robotic peripherals. The protocol steps for the designed high-throughput system, used to purify plasmid DNA from 1.2–1.5 ml bacterial cultures are briefly outlined below.
NOTE: Prior to processing on the high-throughput system, bacterial cells should be cultured in 96-well flat-bottom blocks and harvested by centrifugation for 5 min at 1500 × g in an appropriate centrifuge fitted with a microplate-compatible rotor.
Resuspension and lysis of bacterial cells
Resuspend bacterial cell pellets in 90 μl Buffer P1 by shaking at 1100 rpm.
Add 100 μl Buffer P2 to each sample. Shake samples for one minute at 600 rpm, incubate for one minute and shake again for one minute at 1100 rpm.
Add 500 μl Buffer D3 to each sample. Shake samples for three minutes at 200 rpm and then shake for two minutes at 1100 rpm.
Add 20 μl magnetic bead suspension to each sample and shake for five minutes at 1100 rpm.
Transfer the samples to a round-well block.
Transfer the round-well block onto a 96-well magnet, allow the magnetic beads to separate and then remove the supernatant.
Add 200 μl Buffer DW to each well and shake for one minute at 1100 rpm.
Transfer the round-well block onto a 96-well magnet, allow the magnetic beads to separate and then remove the supernatant.
Add 200 μl Buffer PE to each well and shake for 30 seconds at 1100 rpm.
Transfer the round-well block onto a 96-well magnet, allow the magnetic beads to separate and then remove the supernatant.
Repeat step 9 and 10 twice.
Add 45 μl Buffer EB to each sample and shake the samples for two minutes at 900 rpm.
Transfer the round-well block to a 96-well magnet, allow the magnetic beads to separate and then transfer 45 μl of the eluate to a 384-well plate.
Store the 384-well plate containing eluted plasmid DNA in the cooled incubator.
SYSTEM PERFORMANCE AND RESULTS
System specifications for the high-throughput system we designed are shown in table 1. Sample processing and labware handling steps performed by individual system components were carefully optimized. Up to 5760 samples (60 × 96-well plates) can be processed in a single unattended run, with three 96-well plates being processed in parallel. Sequencing templates can be purified from all 60 × 96-well plates in less than 16 hours; walkaway purification of plasmid DNA and sequencing reaction setup using that DNA are completed in less than 24 hours. Validation studies were performed using pUC19 in E.coli DH10B. Bacterial cultures were inoculated in 1.25 ml LB medium in flat-bottom blocks and grown overnight at 37°C, 250 rpm. Cells were harvested by centrifugation for 10 minutes at 4000 × g in a Sigma 4K15 centrifuge and stored in the Cytomat incubator until further processing. After processing was completed, DNA eluates in 384-well plates were stored in the cooled incubator to prevent evaporation. DNA yields of final eluates were determined using a PowerWaveX Select spectrophotometer (Bio-Tek Instruments, Winooski, Vermont). Ten 96-well plates were chosen randomly. A 20 μl aliquot of each DNA sample was diluted with deionized water to a total volume of 100 μl and absorbance was measured for each sample at 260 and 280 nm. Total DNA yield per sample was then calculated according to the following formula.
Average yields of approximately 2.2 μg DNA with an average 260/280 nm absorbance ratio of 1.73 per sample were obtained, being equivalent to an average DNA concentration of 55 ng/μl.
For qualitative analysis of the plasmid DNA, a volume of 3 μl was removed from each purified DNA sample and separated on a 1% TAE agarose gel. DNA was subjected to electrophoresis for 0.5h at 200 V. Gels contained 0.5 μg/ml ethidium bromide, DNA bands were visualized using an UV transilluminator and an imaging system (Herolab, Germany). Plasmid DNA bands were sized by comparison with Low DNA Mass™ Ladder (Invitrogen Corp., Carlsbad, CA) (Fig. 7).

Gel image analysis of plasmid clones purified using the high-throughput nucleic acid purification system. Bacterial cultures were grown in flat-bottom blocks. Plasmid pUC19 was purified using the MagAttract 96 Miniprep System. 3 μl of final eluates were run on a 1% TAE agarose gel. End lanes are Low DNA Mass™ Ladder size markers.
CONCLUSIONS
The development of novel automated systems for front-end tasks in sequencing projects is driven by constantly increasing sample numbers being processed and cost pressure. The fully automated high-throughput system we designed is ideally suited for laboratories involved in genome sequencing projects. 60 × 96-well plates are processed in a single unattended run, including template purification and sequencing reaction setup, in less than 24 hours. This throughput capacity allows continuous feeding of at least seven ABI PRISM®3700 capillary sequencers or similar instruments capable of processing 96 samples in parallel. DNA yields and purity which could be obtained during validation testing corresponded to the system specifications. Hardware, software, and chemistries are optimized to ensure quick start-up as well as reduced running costs.
