5-hydroxytryptophan biosynthesis

 

I was intrigued by a recent article in the journal Chem (link)  entitled "Creation of Bacterial Cells with 5-hydroxytryptophan as a 21st Amino Acid Building Block" by Chen et al. in the group of Han Xiao at Rice University, wherein they make a strain that metabolically produces 5-hydroxytryptophan for genetic code expansion. It is an interesting example of why metabolic engineering is non-trivial and how scientific research does not progress in a logical fashion.

The genetic code has been expanded for nearly two decades, with a variety of interesting non-canonical amino acids for a variety of different experiments, ranging from precise labelling to long term evolution resulting in fitness benefits for an expanded genetic code. All these experiments involve feeding the cells the non-canonical amino acid, whereas in this paper the cells make it themselves.

In the first few years of genetic code expansion, the field focused on enumerating non-canonical amino acid and finding interesting uses. Then major focus of genetic code expansion became been freeing up space or making new space for the amino acids. Freeing the amber stop codon and overcoming the toxicity of the deletion of its release factor was the major goal of the field as this was a major impediment towards its robust usage.  Recoding a genome is an epic undertaking. Recently in a Nature paper by Jason Chin's group the six serine codons were "compressed" into four resulting in 2+1 codons available for assignment. On a separate front, adding an orthogonal pair of nucleobases (e.g. dNaM and dTPT3/d5SICS) and therefore adding new space for encoding has also proven to be a major minefield that is only now yielding fruit (example). So why was there no effort to make these non-canonical amino acids which seems an easier task?

On a previous blog post I speculate how one could make several amino acids that are similar to existing ones. The major point to note is that it is design on paper: whereas in reality metabolic engineering will present technical challenges. This is true for the 5-hydroxytryptophan paper. On paper 5-hydroxytryptophan seems extremely straightforward to make. Humans make it as a precursor to serotonin, via a single enzyme tryptophan 5-monooxygenase. However, the devil is in the detail and metabolic engineering hits one or more of the following issues:

1. Eukaryotic/archaeal enzymes express poorly in bacteria, even codon-optimised
2. Different taxa require different cofactors
3. The flux of cofactors and metabolites is never right
4. Something becomes toxic

In the case of tryptophan 5-monooxygenase, it is eukaryotic and uses tetrahydrobiopterin, that differs in E. coli. As a result the authors opt for a previous engineered phenylalanine 4-hydroxylase from bacteria. 

Tryptophan is actually very uncommon in protein and in fact in a paper it was eliminate it entirely(ish). So luckily toxicity does not seem to be a problem for the 5HTP paper. But flux is. To overcome this the 5HTP paper adds a cofactor regeneration pathway that boosts production several fold from 10 to 50 µM and the production limit factor is tryptophan. Actually, in an E. coli cell there is only 10–20 µM tryptophan (ref), so toxicity might be a problem, but due to tryptophan depletion. Thus confirming my point that metabolic engineering is capricious but appears simple! This probably also why the paper is followed with the science-y application sections that nobody ever reads in an engineering paper.

Therefore, the question is how many decades be before someone makes a dNaM and d5SICS biosynthesing strain?