Improve Upon Nature: Genetically Encoding Unnatural Amino Acids
27 Jul, 2007 09:49 am
The building blocks of proteins in nearly all organisms are limited to 20 amino acids. It might be perfect for nature, but not ideal for scientists. We have developed a novel strategy to genetically encode unnatural amino acids in mammalian cells, including neurons, for cellular and neuronal studies.
But scientists always look for ways to expand the repertoire of protein building blocks through incorporation of unnatural amino acids into proteins, both in the test tube and inside living cells, because proteins with novel properties offered by unnatural amino acid are of great utility for basic biomedical research such as protein folding, trafficking, and interaction.
After years of effort, unnatural amino acid incorporation has been successfully achieved in bacteria, and then in yeast. The approach mimicked the strategy every cell relies on to incorporate common amino acids into proteins: one stop codon is “hijacked” to encode an unnatural amino acid, and an enzyme is engineered to be capable of sticking the desired unnatural amino acid to a transfer RNAs (tRNA), which is modified to recognize the hijacked stop codon. So basically, a new product line is created for the protein factory and assigned to specifically load unnatural amino acids to the growing protein chain according to the instructions spelled by the genetic code.
Being able to do this in mammalian cells is of tremendous value for scientists, because most biomedical questions relevant to human health have to be studied in the cells of higher organisms and animal models to arrive at the most meaningful answers. For instance, misfolding of the proteins in nerve cells plays a critical role in the aging of the brain and senile dementias, but its molecular mechanism is unclear so far. Unnatural amino acids with fluorescence will be a powerful tool to investigate these questions. However, working with mammalian cells is challenging and far more complicated than yeast or bacteria for many reasons. While it is easy to screen large numbers of mutated proteins in bacteria, the same experiment cannot be done in mammalian cells in the same way. Simply transferring bacterial tRNA genes into mammalian cells has been unsuccessful since mammalian cells fail to produce it. You have to come up with new ways to coerce mammalian cells to do that.
We began by finding gene components that could drive mammalian cells to produce bacterial tRNA - a foreign molecular for mammalian cells that can carry only unnatural amino acids and deliver them to cell’s protein factories. We carefully designed a variety of gene constructs and introduced them into mammalian cells to test whether they could efficiently drive the expression of bacterial tRNA for our purpose. The experimental results turned out that, one of them, indeed worked efficiently as envisioned. Next, an enzyme capable of attaching the bacterial tRNA to the desired unnatural amino acid was needed. For this task we chose an enzyme that was selected from millions of mutated proteins and worked very well in yeast, and transferred this enzyme into mammalian cells. Although yeast is a simple organism, its molecular biology is much more like mammals - usually members from the same kingdom behave very similarly.
After testing our newly developed method in different mammal cells, including nerve cells, we then wanted to apply this technology to solve an otherwise intractable biological question.
Nerve cells transmit their signals by the movement of electrically charged atoms, such as sodium and potassium, in and out of the cells through a “molecular gates” in the cell membrane. Previous studies have shown that when a signal travels along a nerve cell, one of the molecular gates named potassium channel Kv1.4, which belongs to a class of so-called fast-inactivating ion channels, opens briefly and then quickly shuts down. There are two hypotheses about how this may happen. The first proposes that the pore closes when a plug shaped like a ball and chain obstructs the hole. The second suggests that the channel’s flexible head feeds through a small side portal like a thread and blocks the central pore of the channel. Scientists have tried several conventional methods, but none of them has given any insightful clues as to which model was correct.
To address this question, we switched an amino acid in the “thread” domain to a larger one. The idea was that if the ball-and-chain model were correct, a larger “ball” would not alter its effectiveness as a plug; otherwise, it would imply that the thread could no longer go through the side portal or get into the right position. However, even introducing the largest natural amino acid, didn’t reveal any differences. We then performed the experiments with an even bigger unnatural amino acid, an artificial synthesized unnatural amino acid, and found that it really made the difference. The process of inactivation became really slow, supporting the second hypothesis, and indicating the diameter of the flexible head plays a crucial role in the fast inactivation of this channel.
Wang W., et al, Genetically Encoding Unnatural Amino Acids for Cellular and Neuronal Studies, Nature Neuroscience, 01 July, 2007.