A paper published in the Sept. 8, 2006, issue of Science might turn out to be a swan song for hamster proteins.

"The universe of therapeutic proteins falls into two classes: glycosylated and non-glycosylated," senior author Tilman Gerngross told BioWorld Today. Of the two, glycosylated proteins - those with specific sugars stuck to certain amino acids - form the larger group: About 70 percent of proteins need such sugars to function.

To date, the only way to produce most of those proteins in cell culture has been to use mammalian cells, most often hamster cells. Yeast, which is used to produce most non-glycosylated proteins such as insulin, is cheaper and easier to use for protein production. But the innate immune system has evolved to recognize yeast glycosylation patterns, which it usually sees in the context of diseases - that is, infections - rather than cures. As a consequence, it will chomp down immediately on any protein with yeast-like sugars, even if its amino acid sequence is identical to a human one: The half-life of proteins with yeast sugar patterns is "on the order of minutes," Gerngross said.

In Science, Gerngross, associate professor of engineering at Dartmouth College in Hanover and founder, chief scientific officer of biotechnology company GlycoFi Inc. in Lebanon, N.H. (a subsidiary of Merck Inc.), and his colleagues published work showing that they have managed to engineer yeast to add on sugars in a pattern that is both fully human and standardized.

Human proteins are modified in a series of steps; yeast proteins "share the beginning, but then they go off in a very different direction," Gerngross said. Basically, the glycosylation of human proteins proceeds in three stages. In the first step, mainly mannose sugars are added; in a second stage, the sugars will branch out into greater complexity, and as crowning glory, those complex sugars will be capped with sialic acid.

Left to its own devices, "yeast will get stuck in the high-mannose stage," Gerngross said.

His group managed in 2003 to get yeast over the initial hump and into the complex stage. Now, Gerngross said, "we are able to complete the entire pathway with sialic acid." (See BioWorld Today, April 17, 2003.)

The researchers first knocked out the four genes that yeast uses to put its own sugars onto proteins, and then engineered in 14 enzymes that allowed the yeast to put human-type sugars onto proteins. The researchers used their yeast strain to produce recombinant EPO and compared EPO with human-like and yeast-like sugars in rats; while yeast-like sugars led to an inactive protein, yeast-produced EPO with human-like sugars raised hematocrit levels in the animals.

Aside from opening up the ability to produce a number of therapeutic proteins in yeast, Gerngross said the work his team has reported in Science also will improve the production of monoclonal antibodies, which are glycosylated but nevertheless can be produced in yeast. The reason is that monoclonal antibodies have their sugars on the inside, and so the innate immune system misses them in its search-and-destroy mission.

But yeast, at least in the way it is currently used, attaches sugars both differently and more variably than mammalian cells do. And precisely how and where they are attached in turn affects how the proteins fold; in terms of glycosylation patterns, yeast-produced antibodies are actually a mix of drugs, some of which are more equal than others.

"This is not just a matter of humanizing yeast," Gerngross said. Instead, "by being able to control the sugars so rigorously, you can make drugs that are better."