Main

Making proteins in mammalian expression systems is not always ideal, but sometimes it is necessary if you want your protein to contain human posttranslational modifications. The more user-friendly bacterial and yeast systems are not naturally equipped to carry out the multitudes of unique modification reactions. Mammalian expression systems, however, suffer in two major areas; one, that it takes much more time to get sufficient amounts of protein, and two, that it is difficult to uniformly control the posttranslational modifications to make a heterogeneous product.

Luckily, Tillman Gerngross, and his colleagues at Dartmouth College and GlycoFi Inc., did not take this state of affairs for granted. After many years of work, they have succeeded in engineering an artificial Pichia pastoris yeast strain that is capable of performing the entire range of human glycosylation reactions, formerly limited to mammalian culture systems. They achieved this by knocking out four genes coding for yeast glycosylation enzymes and introducing 14 heterologous genes encoding human glycosylation enzymes. This feat was not nearly as simple as it might sound, and as Gerngross explains, “you can't just take a human enzyme and pull it into the yeast because the targeting mechanism will be scrambled...you have to target specific enzymes to the secretory pathway of the yeast in a way where the sequential nature of the reactions is maintained.”

Gerngross and his colleagues are particularly interested in making glycosylated human proteins for therapeutic purposes, so they used their engineered yeast strain to synthesize the heavily glycosylated erythropoietin, an important therapeutic protein for the treatment of anemia. Erythropoietin made in the humanized strain was functional in mice, whereas erythropoietin made in wild-type P. pastoris cells was nonfunctional. “In the past, we've relied on ways of making [therapeutic proteins] that did not afford us the opportunity to control their composition, in particular as it relates to glycosylation,” says Gerngross, stressing that “If you can't control your final composition and you're making a drug that goes into people, that's not a good thing.”

This work could also have a substantial impact on basic research; as Gerngross explains, “with this tool, you can make a protein in its various glycoforms and test them individually for their function.... The community has known for many years that glycosylation impacts activity, but there was no good way of really figuring it out.” With the aid of such cleverly engineered yeast strains, finicky mammalian expression systems may in the future become obsolete.