Just two decades ago we were only beginning to recognize the potential of automated technologies to enhance throughput in drug discovery research. Today, it is difficult to imagine a modern laboratory without robotic equipment.
“We see continuous trends toward increased reliability of robots, partially driven by the introduction of new types of motors,” reports Malcolm Crook, Ph.D., CTO, Process Analysis & Automation (PAA). “Another trend is toward smaller targeted systems, that still flexibly accommodate peripherals as needed.”
PAA will be presenting the “automate.it harmony,” its user interface for software integration, at the Society for Laboratory Automation and Screening (SLAS) meeting later this month.
Conventional small molecule discovery is driven by high-throughput screening (HTS) centers, complete with expensive robotics designed to move microplates between various stations for compound dispensing, assay dispensing, incubation, and optical screening.
The federally funded Molecular Libraries Program supports nine comprehensive screening centers, with the hub at the NIH Chemical Genomic Center—a large-scale ultra-high throughput operation capable of generating over 2 million datapoints per week. The NIH Center was set up to work on lesser known targets implicated in rare and neglected diseases.
“We routinely need to adapt our assays for targets with limited availability,” says Anton Simeonov, Ph.D., chief of the chemical genomics branch. “Working with a unique enzyme or primary cells from a patient with a rare disease calls for conservation of resources at each step.”
The center introduced multiple innovations in designing assays and in adapting them to their fully integrated GNF/Kalypsys robotic system. The system is capable of storing over 2.2 million compound samples, performing multiple assay steps in 1,536-well format, and measuring custom output signals. Three high-precision robotic arms circulate between peripheral units.
“Because we run dozens of protocols, each with its own sophisticated screening logic, our robotic engineers work closely with vendors to modify the source code in real time,” says Dr. Simeonov. The team fully redesigned the traditional dose-response screening protocol to conserve precious starting material. Instead of creating dilution series on the same plate, dilution series are created across plates, meaning that each plate had only one concentration of a given compound. This strategy helps to generate custom data curves and minimizes the potential errors stemming from liquid handling or equipment malfunction.
To further conserve the reagents, the center continuously develops companion assays of matching throughput. “The initial hits typically require confirmation by labor-intensive biophysical and biochemical counterscreens. We miniaturize secondary screens to run in a high-throughput format,” says Dr. Simeonov.
A screen for inhibitors of FEN1, a key DNA repair enzyme, combined a primary fluorogenic screen with a secondary chemiluminescent bead-based assay, both reliably deployed in 1,536-well format. The team continues to change the screening paradigm by adapting other technologies to HT format, including microscale thermophoresis and acoustic dispensing.
Distributed Compound Screening
“High-throughput screening is still a costly and risky endeavor,” says Brian M. Paegel, Ph.D., assistant professor, department of chemistry, the Scripps Research Institute. “Sophisticated robotic systems are expensive to run and maintain, and the compound library capacity does not scale.”
Dr. Paegel envisions a miniaturized, fully automated HTS instrument that can operate in any laboratory. “We are pursuing a new paradigm in compound screening,” he says. “Just like DNA sequencing evolved from being concentrated at a few large-scale sequencing centers to being available at the benchtop and by the bedside, HTS could also become distributed. Distributed sequencing resulted in an explosion of sequencing data of all sorts of organisms. Similarly, distributed HTS would radically transform the drug discovery field.”
To achieve this vision, Dr. Paegel’s team pursues a two-prong approach. First, the method for creating million-compound libraries uses prefabricated building blocks. The resulting compounds mimic natural products. The compound synthesis occurs on beads, and each step of the synthesis is coded in a DNA tag attached to the same bead.
“We are aiming to create a straightforward and robust synthesis process that can be simple enough to execute at the benchtop,” continues Dr. Paegel. “Next, the beads are screened in biological assays using our ultra-miniaturized and integrated microfluidic platform that supplants the need for robotics, microplates, and all of the associated costs.”
The bead library is manually pipetted into this microfluidic circuit, which then separates individual beads into picoliter droplets, already containing all assay reagents. If a compound on a bead produces a positive hit in an assay, the bead is separated from the rest, and the DNA tag is sequenced to determine the synthetic history of this compound.
“We see HTS campaigns of tomorrow to be as distributed and inexpensive as DNA sequencing is today,” concludes Dr. Paegel. The team is currently adapting this technology to screen for new antibiotics and antiretroviral agents.
Industrial laboratory operations depend on robotics to perform routine, repetitive tasks at large scale. But robotics may also be extremely useful in a more dynamic research environment—if only the robots could adjust to a new procedure on the fly.
“It may take longer to program a new robotic procedure via a manufacturer-provided graphical user interface than to go into a lab and pipette yourself manually,” comments Nathan J. Hillson, Ph.D., fuel synthesis division, Joint BioEnergy Institute.
“Moreover, protocols from a given robotics vendor cannot be simply transferred to a robot from another vendor. Because of that, robotics are largely under-utilized in basic and translational research laboratories.” If all experimental protocols were written in the same language, and this language was easily translated into instructions for all laboratory robots, it would provide a tremendous boost for academic science.
Gregory Linshiz, Ph.D., a postdoctoral fellow in the Hillson Synthetic Biology Lab, compares this idea with the way programming languages are used for consumer electronics. A general-purpose programming language, called “C”, maps efficiently to instructions accepted by multiple electronic devices. This capability propelled it to be one of the most widely used programming languages of all time.
“Using this analogy, one can imagine a programming language that maps to all lab automation devices,” continues Dr. Linshiz. “We developed PR-PR (Programming Robots; http://prpr.jbei.org) as an easy, user-friendly, high-level programming language that could interface with any existing robotic platform. Any biologist could learn to program basic procedures in PR-PR in under an hour.”
Furthermore, it is possible to further develop existing software packages to output PR-PR, thus enabling the interface between the software and robotics. The team tested a DNA assembly software (j5) for output of PCR-setup protocols into PR-PR scripts. It takes less than a minute to program a Tecan robot using the j5-generated PR-PR script. The team continues conversion of other protocols into PR-PR to enable greater sharing between research teams.