One day in March of 2006, postdoc Mike Lawrence walked into David Liu's laboratory at Harvard University in a slightly anxious mood. He'd been in the lab for nine months with little to show in terms of good results, and he was hoping this day might turn things around. He had taken on a bold new project with his labmate, Kevin Phillips, to test whether changing the charge of surface residues on a protein could reduce its propensity for aggregating.
Preventing aggregation could be appealing for a number of reasons: understanding neurodegenerative diseases, extending the shelf life of protein therapeutics, and producing better-behaving proteins for lab work such as crystallography. In all of these examples, a protein's propensity for aggregating can wreak havoc on attempts to control its behavior.
Instead of taking the route most scientists might take to avoid aggregation – systematically changing just one amino acid at a time – Phillips and Lawrence went to the extreme, creating a "supercharged" protein with 36 positive charges.
He and Phillips substituted neutral amino acids for positively charged lysines or arginines – a whopping 29 of them. "Just the sheer number of changes introduced into these proteins was pretty massive," says Liu, who was not aware of the design of the project until the results came in. Lawrence and Phillips did hedge their bets, however, creating proteins with random, smaller numbers of mutations in addition to the fully supercharged mutant. (Most proteins fall within the range of -10 to +10.) While scientists had predicted that surface residues could tolerate change and still function properly, the numbers in mind were closer to three or five changes, Liu says with a laugh, "not 30! We didn't anticipate that one could make so many mutations without just obliterating the function."
Such brazen science might usually win a researcher nothing more than wasted time, but not this day. "I came in and looked at the colonies on the plate and there were some green ones," Lawrence says, which meant that the green fluorescent protein (GFP) he had supercharged was folding normally. "But the more likely scenario was something went wrong," he says.
Lawrence sequenced the GFP from the colonies and, much to his surprise, those with the glowing green proteins did indeed have all the charged mutations. "It was a long shot," he says, but it worked, and he was excited to show his results to Liu – who was also excited to see them. GFP with a super negative charge of -30 also glowed, and supercharging also prevented aggregation by the bacterial proteins streptavidin and glutathione s-transferase (J Am Chem Soc, 129:10110–2, 2007). The study confirms earlier work showing that surface charge can discourage aggregation. (J Biol Chem, 280:10607–13, 2005).
The scientists got the idea for supercharging after observing an aggregating-prone protein evolve, to see how natural selection could solve the problem of aggregation. Again and again, the solution seemed to be the same: mutants with the highest charge on their surfaces showed the least aggregation.
While the idea of manipulating charge isn't new, it's "the large amount of charge that's quite interesting," says Harvey Blanch at the University of California, Berkeley. The reason supercharging works is pretty simple. "Basically," says Blanch, "it's electrostatic propulsion between the proteins": Positive charges on a protein's surface keep it at arm's length from the positive charges on its neighbor.
The disadvantages to supercharging, at least in vitro, don't seem to be too great. Lawrence says that the supercharged GFP required about 20% less denaturing chemical to make it unfold, meaning the protein is slightly less stable. The expression levels in bacteria were also lower for supercharged proteins compared to protein with smaller changes in charge. Lawrence concedes that it might be more practical to make just the number of changes necessary to discourage aggregation. Additionally, in vivo, "the tendency to aggregate or be recycled could be important," says Liu.
Fabrizio Chiti at the University of Florence, who published the 2005 paper with similar results, is interested in surface charge for algorithms he's designing that can predict aggregation rates or aggregation-promoting regions. "It's clear charge is an important factor," he says. Whether it is a phenomenon that will work for all proteins remains to be seen. But Lawrence's work supports his hunch, Chiti says: "I think it's a principle."