[curiouscat]

Typically, how much of purified protein would be needed to perform a structure determination of an enzyme? This is a 2000 base pair gene I am looking at that codes for a eukaryotic enzyme.

The final goal is to computationally investigate ways to improve activity, perhaps by swapping specific amino acids. Am I correct to assume that a X-Ray Crystal structure is a necessary pre-requisite to attempt any sort of reasonable computational exercise to improve activity by modifying the active site?

[Yggdrasil]

Typically you would need a few milligrams of protein to begin the process of structure determination.  This essentially requires the ability to express and purify the protein recombinantly from bacteria.  However, obtaining recombinant protein is just the first step in structure determination, and often you will need to optimize the protein itself to aid in crystallization (e.g. determining the domain structure via limited proteolysis experiments, removing unstructured regions etc.).

If homologs of your protein have structures available, you can try homology modeling to estimate what the structure of your protein would look like.  It is also possible to perform directed evolution screens/selections to improve the activity of your enzyme in the absence of any structural information (provided you have a good, high-throughput assay for the activity of your enzyme amenable for screening/selection).

[curiouscat]

It is also possible to perform directed evolution screens/selections to improve the activity of your enzyme in the absence of any structural information (provided you have a good, high-throughput assay for the activity of your enzyme amenable for screening/selection).

Can you elaborate on this part any more? One approach I'm exploring is random mutagenesis by error prone PCR.  Is that similar to what you are suggesting?

What would be a way to develop an assay amenable to high throughput screening? I don't think any assays are available of the shelf. The enzyme is a terpene synthase. The kinetic parameters are known.

The product of the enzyme catalyzed transformation is identifiable / quantifiable via GC / GCMS (it's a terpene ).  The expected yields are in the 50 mg/L range. But that won't help for a high throughput assay, right?

[Yggdrasil]

It's possible to screen activity with GC or GCMS, but throughput would be very low (hundred to thousands of variants per round).  In this case, you would want a more directed approach to generate variants (e.g. by only mutating around the active site or by shuffling various variants of the enzyme) rather than constructing a library through error prone PCR, especially for a ~2kb gene.  Ideally, one could engineer some sort of protein or RNA biosensor to detect the product of the enzyme, which could either produce a fluorescent signal (for fluorescence-activated cell sorting) or directly impact the survival of the organism (for selection experiments).  Of course, engineering a biosensor is a substantial project in itself.

Here are some reviews on various topics you might find useful:
Directed evolution: http://evolve.harvard.edu/111-DirectedEvolution.pdf
Biosensors: http://www.sciencedirect.com/science/article/pii/S0958166916300611

[curiouscat]

@Yggdrasil

What do you think of this paper?

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3969365/

Could I be exploiting this for a high throuput assay for screening my terpene synthase? Thoughts?

[Yggdrasil]

Yeah, that could works as a good way to screen different mutants, especially if you don't have the instrumentation to do a high throughput GC-MS screen.  One disadvantage is that you're looking at disappearance of the reactant, so if you have an enzyme that catalyzes a variety of reactions, you can't optimize it to produce a specific product over the other side products.  A good workflow might be to use the colony color screen as an initial step to remove non-functional mutants from your library and then use GC-MS to look at the enzymes from the white colonies in more detail.  Still, because the paper reports only screening through thousands of colonies, the throughput of the screen is still going to be relatively low, so you'll need to carefully consider how you generate diversity in your library to give yourself the best chance of finding a good hit.

[curiouscat]

One disadvantage is that you're looking at disappearance of the reactant, so if you have an enzyme that catalyzes a variety of reactions, you can't optimize it to produce a specific product over the other side products.

True! I never thought of that bit.

Even in my case, the same enzyme results into a mixture of related terpene isomers.  I was focusing on yields in aggregate, somehow taking as a given that the product distribution doesn't change much.

Well, you are right, once I start changing the residues within the enzyme this assumption no longer holds. I cannot take that as a given.