What do you get when you have a molecular biologist with a penchant for robots? If you're lucky, you'll get Stephen Hughes, a research molecular biologist at the USDA Agricultural Research Service (ARS) National Center for Agricultural Utilization Research (NCAUR) in Peoria, Ill. Hughes was in the pharmaceutical industry before joining NCAUR nearly four years ago. In bringing his pharmaceutical background to agricultural research, he also brought one of the most effective tools for genomic research.

In collaboration with the Hudson Control Group Inc. in Springfield, N.J., Hughes and his team developed a plasmid-based proteomic workcell that integrates all of the molecular, microbiological and biochemical techniques used for high-throughput strategy into a single robotic platform. The robot will ultimately be used for identifying important xylose-utilization genes and for improving strains of yeast that ferment xylose to ethanol.

Hughes is part of NCAUR's bioproducts and biocatalysis research unit, which is focused on developing practical, useful catalysts and methods for making ethanol from cellulose. The broad goal of the unit is to develop new microorganisms and biocatalysts that can be employed in the fermentative conversion of renewable agricultural materials to fuels and other value-added products. The research entails engineering existing fermentative microorganisms to possess desirable traits for industrial fermentation of lignocellulosic material, or searching for new microorganisms that possess these traits.

This research project has been ongoing since September 2004. Hughes' specific objective is to create efficient xylose-fermenting Saccharomyces cerevisiae strains (a known ethanologen). His approach is the application of high-throughput screening procedures to develop, by use of directed evolution and gene shuffling, microbial strains and enzymes with superior ability to convert agricultural materials to biofuels and bioproducts. The robot can screen tens of thousands of candidates for improved variants.

High-throughput screening is a combination of robotics, data processing and control software, liquid handling devices and sensitive detectors. These technologies are used to screen large numbers of compounds rapidly and in parallel order to collect a large amount of experimental data in a short period of time. High-throughput screening has been used in the pharmaceutical industry to quickly develop drugs. It is effective because each experiment is controlled and repeatable, and the results quickly weed out the ineffective traits.

Until the NCAUR robot, no integrated robotic platform was available to carry out the entire process from creation of plasmid libraries to expressing cloned genes and, finally, to functional testing of expressed proteins. The NCAUR robot takes high-throughput screening where it hasn't yet been—full automation. The robot extracts genetic material, makes DNA copies of these genes, clones and cultures the copies and creates "cDNA libraries" from these plasmid clones. From these libraries, the robot can pick the genes with a favorable response and sequence their nucleotide bases. Next, the robot identifies the corresponding protein for the gene and inserts the desirable genes into Saccharomyces and other yeasts. In other high-throughput screening applications, some of these processes would still be done by hand. Because of the speed and accuracy of the robot, it can perform these tasks hundreds or thousands of times faster than a human, according to the ARS. What the robot does is routine molecular biology, Hughes says, but this is the first time it's been fully automated. "This is the only high-throughput, high-content screening machine out there," he says.

Hughes and Philip Farrelly, president of Hudson Control, had several pieces of the robot custom designed and built specifically for this application. The combined technology is now valued at over $1 million, Hughes tells EPM. It took two years to develop the parts, he says. Many of the parts came from different research fields and were integrated together within the unit. The scheduling software, which was developed by Hudson Control, is the key to the integration. The software also provides for full automation because the database of genes is built in.

When he first started working with robots, they were faster than the computers used to run them, Hughes says. "Actually, that's where some of the sophistication in our software came from," he says. "They had to write the software so it would slow down enough so you could do the robot schedule but you could wait for the computer to catch up."

Test Project
Producing yeast optimized for cellulosic ethanol production is the primary goal of the project, but the robot was tested for operational protocols with a project that involved moving a protein from the wolf spider into yeast. The protein is being considered as a natural insecticide for corn earworms and fall army worms. The gene for the protein was mutated into thousands of variations. Hughes and his team then used the system to add all those versions of the gene to a variety of brewer's yeast, and screen the resulting strains for production of the spider protein and its ability to kill corn ear worms.

Hughes used the system to perform a high-throughput screen of yeast clones to find optimized cellulase F genes with improved pH and higher temperature stability. Each plate in the system has 96 wells, and Hughes grew eight clones of the yeast in each well. He says this first test was a "brute force" test that analyzed more than 23,000 variations in the cellulase F protein. The tests found several varieties that showed a higher activity cellulase F variation of the desired protein that was active at a lower pH (valuable because industrial fermentation is usually done below a pH of five) than strains they identified by manual methods.

This process utilized a picking strategy that multiplexes, or combines several mutagenized clones in each well, to reduce the amount of reagent required. The robot makes possible fully automated multiplex picking routines followed by automated plasmid preparation and in vitro expression of proteins. The proteins with desired characteristics are then identified and isolated.

Yeast
Now that the system is in place and tested, Hughes is gearing up to process yeast for cellulosic ethanol production. By August, all components are scheduled to be in place for a full operation. According to Hughes, the target organism would be a yeast (or other organism) that expressed four or five different enzymes that would break down cellulose and starch. The yeast's metabolic pathways would be adjusted with the corrective genes so that it could perform simultaneous saccharification and fermentation. Basically, the enzyme would be included in the yeast so that the conversion to sugar and the fermentation into alcohol could occur by using one organism. In addition to ethanol production, Hughes aims to include other enzymes that could be expressed after ethanol is produced. He envisions that the product leftover from ethanol production will have several enzymes that could be used to create different products. These products could include pesticides, building blocks for biodegradable plastics or anti-cancer compounds.

The robot was designed to create a target organism. Saccharomyces was the organism of choice for several reasons. Firstly, it's a known ethanologen, meaning that it is already successfully used in ethanol production. Secondly, the entire yeast genome with 5,632 genes has been fully sequenced. Thirdly, Hughes partnered with Josh LaBaer, director of the Harvard Institute of Proteomics, who has a collection of 80,000 Saccharomyces organisms in his library. "We've created a whole system where we can add entire libraries into the yeast and move them quickly in and out," Hughes says, noting that he will start with LaBaer's library of genes from Saccharomyces cerevisiae.

Despite all this work, Hughes is quick to point out that yeast may not be the answer. "This may not be the fix, but this may give you insight into what you need to do," Hughes says. "We're interested in finding another system, whether that's going to be a new yeast or another microorganism that makes ethanol that can get us into the future."

In May, Hughes co-authored a paper for the peer-reviewed journal Proteome Science. In its description of the project, the paper claimed that this research opened the door for a new kind of proteomics (study of proteins). It was the most read article of the year for that journal, Hughes says.

In the annual report for Hughes' research unit, the robot is listed as the single-most significant research accomplishment during the 2006 fiscal year. The robot "represents an advancement in the field of laboratory automation," according to the report. "The robotic integration of microbiological, molecular, and biochemical techniques will facilitate the development of improved biocatalysts for converting biomass to fuel and valuable products. The workcell will also find broad application in other fields of agricultural and pharmaceutical biotechnology."

Joseph Rich, research leader for bioproducts and biocatalysis at NCAUR, agrees that this project has far-reaching potential. "This is a technology that is completely independent of whatever area you want to work on," he says. "It works today on ethanol. Tomorrow it could work on cancer peptide, on butanol or pesticides—it could work on anything. Even after the ethanol questions are solved, we'll be cranking on the next problem. There will never be a lack of something to look at. It will just be on to the next one, and we'll just get better and better at doing it."

Anduin Kirkbride McElroy is an Ethanol Producer Magazine staff writer. Reach her at amcelroy@bbibiofuels.com or (701) 746-8385.