© Alfred Pasieka Photo Researchers, Inc.

Gadget freaks love to "mod" their toys, and protein engineers are no exception. Just as people have found ways to trick-out their iPods, protein engineers have devised strategies to "tune up" enzymes, figuring out how they work and conferring new properties in the bargain.

Until recently, though, researchers were limited to the 20 naturally occurring amino acids and site-directed mutagenesis. But advances in chemical biology and genetics are providing access to the much larger realm of "unnatural" amino acids – those that can be chemically synthesized but seldom or never occur in nature.

"The potential for adding new properties is endless," says Yi Lu, who studies engineered metalloproteins at the University of Illinois at Urbana-Champaign. Researchers are using amino acid analogs to tease out the roles of individual residues in protein function, incorporate fluorescent labels for protein-trafficking studies, add photoactivatable crosslinkers for protein-protein...


Lu and others point to Peter Schultz of The Scripps Research Institute in San Diego, Calif., as a major pioneer in the field. Since the 1980s, Schultz has been developing genetic methods to incorporate novel amino acids into proteins, based on the redundancy of the genetic code. In the earliest iterations, Schultz used pioneering methods developed by Sidney Hecht of the University of Virginia to chemically charge the transfer RNA (tRNA) that recognizes the UAG (amber) stop codon with an unnatural amino acid. Schultz could then force translation machinery to insert the modified amino acid whenever it encountered an amber codon, rather than terminating protein synthesis, in test-tube reactions.

This method did not catch on widely, notes Lu, because the chemical charging reaction was difficult, and protein yields from in vitro translation tend to be low. But Schultz has since refined the approach.

Each tRNA has an enzymatic partner called an aminoacyl-tRNA synthetase, which recognizes the tRNA and charges it appropriately in vivo. Schultz mutagenized genes for the amber suppressor tRNA/synthetase pair from the archaeon, Methanococcus jannaschii, to make them recognize desired amino acids and to ensure they did not interact with other tRNA/enzyme pairs. He then introduced plasmids encoding these genes into Escherichia coli, along with another plasmid containing the amber-mutated gene for the protein to be translated. Shigeyuki Yokoyama of the University of Tokyo has used a similar approach to introduce unnatural amino acids into proteins in mammalian cells.


Courtesy of Peter Schultz

By creating new components of the protein-synthesis machinery Peter Schultz and coworkers at the Scripps Research Institute have added a large number of amino acids with novel physico-chemical and biological properties to the genetic codes of both eukaryotic and prokaryotic organisms.

The advantage of this new system is that it both eliminates the need for chemical charging, and improves yield by allowing the protein to be made in whole cells rather than in vitro. The cells tolerate this alternate use of the amber stop codon quite well, explains Schultz. "Very few genes terminate in an amber codon," he says, adding that bacteria also generally have multiple stop codons that would largely prevent unwanted read-through of other proteins.

In the last two years, Schultz has used variations of this system to incorporate 5-hydroxy-L-tryptophan, a fluorescent amino acid that can also be used as an oxidatively activated protein crosslinker,1 and beta-N-acetylglucosamine-serine, a preglycosylated residue that demonstrates the possibility of making therapeutic proteins that are more faithful replicas of human variants.2 He has incorporated amino acids with novel reactive groups, such as alkynes, that can be used to add fluorescent labels or glycosyl groups following translation.3 And he has extended the method to both yeast and mammalian cells.

"We've sent out a huge number of strains that put in photo crosslinkers," says Schultz. "Those, together with the addition of small fluorescent amino acids and those that can put in glycosylated amino acids are the ones that people are really attuned to."

In 2003, Schultz founded a San Diego-based biotech called Ambrx to apply his technology in the drug development arena. Protein drugs like human growth hormone (hGH) are plagued by problems. They can trigger potentially dangerous immune reactions, and they tend to be vulnerable to proteases, leading to a short half-life. One solution is to shroud them in hydrophobic polyethylene glycol (PEG) molecules, which shield them from both antibodies and proteases. But the typical, chemical "PEGylation" process produces a mixture of differently modified proteins with varying clinical efficacies.

Company scientists used Schultz's methods to produce an hGH variant that contains p-acetylphenylalanine residues at specific positions. PEGylation at these residues produced a uniform population of highly bioactive hGH, says Ho Cho, Ambrx's director of molecular technology and process development. The molecule stays in circulation 10 times longer in animals, raising the possibility of weekly injections rather than the daily injections currently needed, says Cho, who expects the molecule will enter clinical trials in the latter part of 2006.


Thomas Magliery, who studies protein structure at Ohio State University, was a member of Schultz's lab when the tRNA/synthetase system was developed. "The method is coming into its own," he says. "It's very simple for unnatural amino acids that already have synthetases made for them."

But the method has an "Achilles' heel," he adds: A new synthetase must be developed for each novel amino acid, a process that can take months. Moreover, while it is relatively simple to make a synthetase specific for a very unusual amino acid, it is more challenging to make one that recognizes subtle differences between amino acids. But these subtle differences are what interest protein engineers like Lu.

Lu seeks to understand how proteins bind metals, and to design new biocatalysts for drug synthesis. Such catalysts could be used to produce pure enantiomeric drugs (as opposed to the racemic mixtures chemical methods usually produce). They might also bind metals not used in natural proteins, potentially expanding their synthetic capabilities.

Lu uses an alternate to Schultz's method, expressed protein ligation (EPL), to make modified metalloproteins. In EPL, a protein produced in vivo by conventional methods is chemically ligated to a synthetic peptide containing the desired unnatural amino acid or acids.

"The biggest advantage [of using unnatural amino acids] is that you can separate one factor from another in designing proteins," Lu says. "You can change the electrical factor without changing the steric factor." By designing the right substitutions, it is possible to achieve almost "atomic level replacement" of side chain atoms, he adds.

Massachusetts Institute of Technology molecular biologist Uttam RajBhandary, and research scientist Caroline Köhrer, have devised another approach.4 The pair couple unnatural amino acids to suppressor tRNAs chemically, in vitro, and then import them into the cell. "It's completely general," explains RajBhandary, "in the sense that once you have identified an appropriate suppressor tRNA, you can attach any unnatural amino acid to it and import it into the cell, where it will deliver the unnatural amino acid to the protein at the site you want."


Novel amino acids can also be useful in the design of small peptide drugs. "Peptides are the most common form of ligand in all of biology," says Gregory Verdine, professor of chemistry and chemical biology at Harvard University. But peptides are difficult to develop into drugs because they can unfold from their biologically active conformations, and are often too hydrophilic to cross the cell membrane.

"Peptides are on the outs" as drugs, because of their stability and permeation problems, says Verdine. Unnatural amino acids, he asserts, are "widening out the palette of molecules that can be called a drug."


© 2005 Elsevier Inc.

An overview of Hiroaki Suga's column-based aminoacylation system, which incorporates nonnatural amino acids into proteins with an RNA-based aminoacyl tRNA synthetase analog, or ribozyme. (From D. Kourouklis et al., Methods, 36:239–4, 2005.)

Verdine collaborated with late-Dana-Farber Cancer Institute researcher Stanley Korsmeyer to engineer a novel anticancer peptide called SAHB, for "stabilized alpha-helix of BCL-2 domains."5 SAHB mimics a portion of the apoptosis promoter, BCL-2. Verdine engineered the peptide to contain novel amino acids bearing hydrocarbon chains that could be linked together chemically, in effect "stapling" the peptide into an alpha-helical conformation. This conformation improves the peptide's stability, its ability to cross the membrane, and its intracellular bioactivity.

More to the point, the peptide suppressed leukemia in mice. Verdine and former Korsmeyer postdoc Loren Walensky plan to initiate pre-clinical studies in the near future under the aegis of Harvard's new Program in Cancer Chemical Biology, which Verdine directs.

To have many more successes like this, drug developers will need whole libraries of peptides containing unnatural amino acids that they can screen. Hiroaki Suga of the University of Tokyo is developing methods to make such libraries. He has developed a 45-nucleotide catalytic RNA, or ribozyme, that can charge tRNAs with unnatural amino acids. The system can incorporate multiple unnatural amino acids per peptide and imposes no restrictions on which specific amino acid is used.

Schultz's methods are "really impressive," says Suga, but "there are things they cannot do very well. We are moving in a different direction. This is an enabling technology for producing peptide diversity." Suga currently is in discussions with an RNA-synthesis company to produce the ribozyme commercially.

Chuck Merryman and his mentor Rachel Green at Johns Hopkins University have developed an alternate strategy to produce libraries of peptides containing N-methylated amino acids. "N-methylated peptides are known to be membrane-permeable and resistant to proteases," says Green, making them good drug candidates.

Merryman's idea was to N-methylate the amino acids after they were charged onto tRNA's, eliminating the need for Schultz's complex genetic manipulations; the charged tRNAs are then used to make peptides by in vitro translation of a random mRNA library. "The trick was to find the right chemical conditions," says Green, adding that Merryman's conditions can now reliably methylate 15 of the 20 naturally occurring amino acids.6

"The dream is to make 1013 different molecules in a single experiment," says Green. This random peptide mixture would be passed over affinity columns to identify potential inhibitors of HIV protease, for example. Peptides that bound would already possess favorable stability and permeability characteristics, making them more attractive than unmodified ones.


Biologists who are studying protein trafficking or protein-protein interactions generally do not need comprehensive incorporation of novel amino acids. For these studies, they find amino acid auxotrophs more useful. Auxotrophs are strains of cells – whether bacterial, mammalian, or yeast – which lack the ability to make a certain amino acid. Since they must get that amino acid from their environment, such cells will take up and incorporate unnatural amino acids that closely resemble their natural counterparts.

Christophe Thiele of the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden, has developed photolabile analogs of leucine and methionine that cells will substitute for natural leucine and methionine between 10% and 20% of the time.7 When activated by light, these analogs produce covalent crosslinks with nearby proteins.

Thiele is using these compounds to study large protein complexes that regulate cholesterol biosynthesis in the endoplasmic reticulum. "Many events in membranes are regulated by large protein assemblies," he says. "We are aiming to define and resolve these interactions with covalent evidence." Thiele has offered these analogs to the research community and received almost 40 requests. "There will be many, many applications," he says.

The development of new methods for incorporating unnatural amino acids into proteins has clearly sparked renewed interest in using them. But Lu points out such proteins are still "not trivial to make." Complete chemical synthesis is limited to peptides of less than 100 amino acids, and is very expensive. Both native chemical ligation and EPL require a cysteine at the junction between the two peptides, and make it difficult to incorporate unnatural amino acids a long way from the N- or C-terminus. And auxotrophs are not always available, particularly in organisms like yeast that have multiple redundant biosynthetic pathways, says Thiele. "Although the potential for this is enormous, very few labs are really working on it," concludes RajBhandary. "First, you require very specialized knowledge, and second, it takes a lot of effort. That is why we are working on approaches that are of general use."

"We can demonstrate the power of this method when we can show how to put [unnatural amino acids] into proteins in high yield and at reasonable cost," Lu says. That time may not be far off: Ambrx is currently in talks with a "major research tool" company to make Schultz's strains and reagents commercially available. Tamara Hendrickson, who studies protein translation at Johns Hopkins University in Baltimore, says, "We're approaching the time when there will start to be kits."

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!