The cow rumen can hold over 150 liters of digesting food—giving the researchers plenty of material to work with. Cows are particularly advantageous test subjects for this type of study for another reason. Agricultural scientists have developed a system for putting a window-covered opening into cows’ rumens. It’s literally possible to look into the stomach through this window, called a fistula, and to put experimental samples inside and then draw them out. Working with researchers at the University of Illinois who have cows fitted with fistulas on campus, the Joint Genome Institute researchers put bags of switchgrass into the cow stomachs, let them sit for 48 hours, and took them back out again. Microbes found adhering to the switchgrass were presumed to be involved in fermenting it.

The researchers then separated out the microbes, broke them down, and sequenced all the genetic material they found. The researchers were able to produce a tremendous amount of data about the genes, and some of the genomes, found in the cow rumen. They found about 250 billion base-pairs worth of genes, about 10 times more than make up the human genome.

The challenge was then to interpret all that data. Using the high-performance computing facilities at California’s Lawrence Berkeley National Lab, the group compared the cow-microbe sequences with a database of sequences known to code for enzymes that break down carbohydrates. This led to a pool of 28,000 genes for further study. The researchers then used lab bacteria to make the proteins coded by 90 of these genes, and tested their functionality. About half of them were able to break down cellulosic materials.

In addition to the 28,000 genes identified, the researchers were able to reassemble the genomes of several previously undiscovered microbial species. To test this part of the study, they isolated a single unculturable bacterial cell from the rumen samples, and sequenced its genome. It matched one of the ones they had assembled. “This biological reality check makes us quite confident,” says Rubin.

How the new database will be used isn’t clear. “It’s an encyclopedia for people to pull out,” says Rubin.

Previous efforts to come up with new cellulytic enzymes haven’t resulted in much. There are two ways to do it. One is to try to make the enzymes more active. Another way, one that Rubin and Caltech’s Arnold both consider more promising, is to find or make enzymes that are not only more active, but also work under extreme conditions that might help facilitate the breakdown of tough plants, such as high-temperature, high-salt solutions–conditions that destabilize today’s enzymes.

Meanwhile, researchers are applying metagenomic analysis to other microbial communities that ferment tough plants. David Weiner, head of research and development at the enzyme company Verenium, says his company already has termite-microbe enzymes in its product-evaluation pipeline. The company was involved with the earlier termite studies done by the Joint Genome Institute researchers and has developed a platform to speed up the testing of new enzymes. Weiner says the company is also looking for enzyme genes in other ruminants, including zebras, and in samples taken from, for example, decomposing logs. Verenium sold part of its celullolytic enzyme business to BP last fall.