December 1, 2007 (Vol. 27, No. 21)
Flexible Systems and Intelligent Automation Are Key to Cost-effective Implementation
The ELRIG/SBS “Drug Discovery: The Application of High Throughput Technologies to Target and Lead Identification” conference, which was held in October in Nottingham, U.K., featured a special section focused on the challenges of sample management. Some of the topics included collecting and processing biological samples, adapting storage facilities to handle new types of compounds, software management, and maintaining sample integrity.
One of the biggest issues in software management is finding hits within large screens of considerably sized sample collections, said Kevin Moore, marketing manager, Tecan Group (www.tecan.com). “There’s definitely a trend away from doing very large sample collection screening because it’s quite expensive. In the past, people worked in a plate format, which meant a few hits on each plate—not very cost effective. Most researchers now use single-use tubes. When you have a one or two million sample collection, it is difficult to perform analytical studies on every sample.”
Although the storage systems are currently adequate, improvement is needed in the way laboratory information management systems (LIMS) communicate with robotics for screening and software management.
“The interfaces are difficult because each company has its own LIMS. I see incremental changes in this area, as opposed to huge advances,” Moore stated. “The improvements are going to be in how the intelligence comes in to actually cherry-pick the collections to make sure what you screen is a more relevant screening file. There is a movement to try and use computer-based methods to prescreen samples.”
Pfizer (www.pfizer.com) embarked on building a centralized automated storage facility for its compounds in 2003. “It’s the Amazon.com era for all of us now; you want your researchers to have a web-based application where they can go to a shopping cart and get what they need and have it in their lab quickly,” explains Diane Johnson, associate director. Since the company had already built a center for its compounds, the basic design for its biofluid sample storage was already established.
The company had a large robotics storage facility for compounds stored at -20°C, but there were some novel challenges around the -80°C storage required for biofluids. “When you are physically pulling samples from a -80°C freezer to a
-20°C freezer environment, there is a temperature flux, and that’s a huge delta,” says Johnson. “Our challenges were minimizing that temperature flux, the condensation issues of working in New England, and the transition of samples in and out of the freezers.”
The company partnered with REMP, a Tecan Group company, to develop software to accommodate picking an order across multiple storage racks and across multiple freezers.
The robotic freezers monitor the upper and lower operational limit and will close the freezer door when the freezer is warming outside of the set limits. “Any given rack for any different sample order has to be modulated by those temperature constraints; we’ve set a five degree delta. REMP built complicated algorithms for the robot to be able to go to another freezer, pick a sample, go back to a different freezer, open the door, and continue,” notes Johnson. Software flexibility enables users to tighten or loosen the tolerances according to throughput.
Another challenge was the migration of legacy samples. Some samples were received in different size tubes with different tops, variable volumes, and assorted concentrations. This required the development of robotics to hold different tube types, different cappers and decappers, and handle variable volumes. Biofluid samples are defrosted once; multiple replicates are pre-aliquoted and placed in sealed tubes at -80°C. When the customer orders them, they are in single-use tubes, which helps maintain their integrity.
Amphora Discovery (www.amphoracorp.com) recently shifted its focus as an internal drug discovery company to a screening services company. The company originally assembled a flexible compound-management system because its technology was a bit different from the standard technologies being used, explained John K. Dickson, Ph.D., director of chemistry and compound services. “We knew that, as we grew, we would need to supply compounds to a variety of different internal organizations such as HTS, cellular assays, and analytical chemistry as well as for interactions with outside companies.”
When the company began focusing on providing services, there were a lot more compounds coming in. Since the original system was flexible, said Dr. Dickson, “there were only a few aspects that we needed to adapt to make it a smooth operation.”
Liquid-handling methods were modified for more flexibility and most of the processes automated to remove human error. In order to minimize the number of freeze/thaws once a compound goes into solution, a single-use plate system is used so that the compound and data maintain integrity. Most of the software was developed in-house; proprietary modules for tracking amounts, container types, and plates make sure all manipulations are recorded, according to Dr. Dickson.
“We were forced to have flexibility up-front and to make sure we were able to handle a variety of things,” added Dr. Dickson. “We can handle all forms of compounds from the customer including dry solids, dry films, 96-well plates, solutions. The bottom line is that data quality is not influenced by compound management.”
Researchers at Cephalon (www.cephalon.com) were faced with limited storage space, so they had to develop solutions to economize space. They also wanted to implement a process to prepare assay-ready plates for the end user during the night in order to have them ready for high-throughput screening the following morning.
“We were looking for automated equipment that would handle the compound plates and dispense the compounds onto the plates without intervention by the user,” said Jean Husten, Ph.D., senior scientist, lead discovery and profiling.
The TekCel automated plate management system from Hamilton Storage Technologies (www.tekcel.com) was chosen a few years ago, and the company integrated a tool into the bench deck that senses the buffer and then compounds are pin-tooled into the plate. Additional equipment is used to reformat compounds from 96-well plates to 384-well plates. Since the TekCel bench has a nitrogen environment option, it allows compound processing in an inert environment, which is important for preventing compounds from oxidizing, reported Dr. Husten.
Another compound-dispensing delivery system recently implemented at Cephalon is the StoragePod™ system (Roylan Developments; www.spod-technology.com), which is filled with nitrogen, providing a dry and oxygen-free environment. “Compounds are kept in 100% DMSO; keeping the water content down helps to prevent precipitation of the compound,” said Dr. Husten.
StoragePods are also being used to store archived sets of compounds off-line from the TekCel Plate Stores. Dr. Husten’s group has screened more than 10,000 plates, with good confirmation rates from the positives received via high-throughput screening. She will be purchasing a tube storage handling system to create more focused libraries for screening against particular targets and is considering various software solutions for sample tracking.
As an emerging biotechnology company, Lectus Therapeutics (www.lectustherapeutics.com) has unique requirements for designing its compound-management system. “We need a flexible system to grow with our collection, while maintaining a cost-effective investment in infrastructure,” said Adrian Kinkaid, Ph.D., head of assay development and screening.
Management for Growth
In addition, the company wants to retain the option to screen compounds at higher concentrations in order to facilitate construction of a structure-activity relationship for a target molecule. This requires storing compounds in DMSO at room temperature under nitrogen, avoiding freeze/thaw cycles. “This allows us to minimize the risk of compound precipitation at lower temperatures and through phase changes,” explained Dr. Kinkaid.
“The most challenging part of the system design was the nitrogen cabinet, which is used to dissolve compounds in dry DMSO and to reformat mother stocks into daughter plates under an inert atmosphere.” The cabinet had to be large enough to house liquid handlers but small enough to enable rapid flushing with nitrogen. Compounds stored in StoragePods can be flushed with dry nitrogen or argon to 0.5–2% oxygen and sealed with a positive pressure of 1 psi. Easy access is achieved via a removable front panel.
One requirement not being met, said Dr. Kinkaid, is reliable in silico predictors for compound solubility in DMSO and water, which would allow for selection of appropriate compounds/libraries for enhancing the company’s screening library.
UK Biobank (www.ukbiobank.ac.uk) is a large prospective study investigating the role of genetic, environmental, and lifestyle factors in the causes of major diseases of people between 40 and 69 years old. Data and samples (urine and blood) are being collected from 500,000 participants at different locations.
Pilot studies concluded that blood samples must be processed and stored in low temperature archives within 30 hours after obtainment. “This data really set the entire sample handling and storage infrastructure,” said Tim Peakman, Ph.D., director.
The other important factor is scale; each participant will give seven different vials of either blood or urine. Upon processing, one individual will produce 28 1-mL, 2-D bar-coded tubes. “When you multiply that through, we’ll be producing about 20,000 1-mL aliquots per day, which will be split between two low-temperature archives,” noted Dr. Peakman. “Approximately two-thirds of the samples will be archived in a
-80°C fully automated sample-handling archive. The remaining one-third will be kept at a more manually controlled site in liquid nitrogen. The two archives are geographically separate as a security measure.”
The initiative’s approach in building a high-throughput blood and urine processing facility and storage archive “has given us enormous benefits, significant cost savings, and allowed us to cope with some quite significant changes to the protocol,” reported Dr. Peakman.
For example, the selection of tubes was changed in the original protocol based on data from stability studies. “Handling different tubes via automation was standard practice but not the ability to discriminate the layers of fractionated blood with high accuracy and convert that into liquid handling instructions. That’s what all our efforts went into at the beginning of the design.”
In order to track samples, Dr. Peakman says they worked with several tube suppliers to have the tubes bar coded in a special way. “This enabled us to do two things: make an anonymous link to the participant and their samples, and it enabled the system to discriminate exactly what type of tube it was processing.”
Total design and building of the system took about two and a half years. Samples are continuously monitored for quality and stability, and the group is now in the process of setting up some long-term stability studies of archived samples.