Recently I was reading about drug discovery.

As far as I know, there are currently 2 sources of creating new compounds by:

  • Screening from natural sources.

  • Synthesizing by add/remove/replace-ing the functional groups of a known compound using chemical processes.

On the other hand, molecular pharming is the process of producing drugs by integrating the drug-encoding genes into bacterial (typically E. coli) genome.

I came up with the idea that if we create mutations in these drug-producing colonies, by UV light for example, eventually by chance we can obtain some new compounds once the mutations occur in the drug-encoding region or related genes.

I want to search in the previous studies for those who has come up with this idea already, I tried with several keywords and there's no luck.

I would want to ask is this idea practically possible? How high is the chance of having new compounds compare to screening natural sources and chemical systhesis? And if someone has already done this study, can you please provide me some references on this matter?

  • $\begingroup$ Would you consider the idea of site-directed mutagenesis for creating mutations? $\endgroup$ Jan 11, 2017 at 16:14
  • 1
    $\begingroup$ It could be site-directed mutagenesis or any kind of random mutation when exposed to stress. But the product is unknown. Because the use of known mutant strain is not the idea at all. $\endgroup$ Jan 12, 2017 at 5:34

2 Answers 2


I'd say this is possible, but not really practical. Mutating genes by UV light is not that hard, it's just the screening for the (new) compounds that makes it difficult. You would probably be better of creating diversity using established chemical methods.

Usually UV mutagenesis is done for whole organisms, but some people probably also tried this for single genes (even though they'd hit the whole genome). This is the easy part; you could get an enormous amount of diversity with very little work. Even though mutations that change enzymatic products are probably much more rare than mutations that change substrate specificity or product chirality, people find those mutations (example from the top of my head: p450 hydroxylation patterns, although they probably used targeted mutagenesis).

Determining if a specific compound is produced by an organism is also not very hard. You break all the cells, put the solution on a column and check with your favorite analytical method whether or not your compound is in there. Checking for new compounds is much harder. You need better analytical methods such as MS/MS as you don't have analytical standards, and the adapted molecule might not even be detectable with the analytical method you used for the starting molecule. We are very good at detecting needles in haystacks, but determining if there is a needle, a stick, or maybe a piece of hay from another species in the haystack is much harder. You'd already have to predict what kind of altered molecules you expect to find.

When you have successfully done all this, you have identified a single new molecule. Nowadays, people use libraries with >100,000 compounds to screen against their disease protein of interest.

The idea of mutating enzymes to produce new products is not that strange, people work on this, publish, find a lot of interesting things and we learn a lot about enzymes. As an approach for drug discovery this approach is just not high-throughput enough.


Indeed, small molecule and secondary metabolite detection has been, and often still is done through both random and rational mutagenesis. UV light is one option, but often it's easier to use slightly more aggressive methods such as ethanemethanosulphate (EMS) exposure.

Streptomyces for example, the model organism for secondary metabolite production has been subjected to all manner of mutagenesis methods throughout the years (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3472513/).

A random mutagenesis approach used frequently these days is genetic rather than chemical, and uses the mobile elements transposons for random insertions. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3416362/)

Much secondary metabolite production these days is done via modification of Non-Ribosomal Peptide Synthase clusters (NRPS). These clusters exhibit something called the 'colinearity rule' (http://www.sciencedirect.com/science/article/pii/S1074552110000499) whereby, the 'modules' that make up the protein complex, play 'pass the parcel' with the growing seconday metabolite, with each module subsequently adding a new moiety - like a conveyor belt. It's possible to therefore predict the molecule an NRPS will make just from the modules and mutations you see within it, whether produced randomly or rationally.

However, it's often the case that we understand the myriad of secondary metabolite molecules that an organism is capable of producing so poorly, that simply screening the wildtype organism turns up hundreds of thousands of molecules before you even need to consider modifications.

  • $\begingroup$ I think your sources point more towards mutagenesis to produce more of a certain molecule, not towards producing completely new molecules. $\endgroup$
    – VonBeche
    Jan 13, 2017 at 16:53
  • $\begingroup$ I hadn't interpreted the OPs question in that manner, but on rereading I guess that's possibly what he meant. In that case I would say random mutagenesis to produce totally novel compounds is functionally impossible. On average, mutations are neutral, and of those that aren't they're more frequently deleterious than advantageous so you'd need an extremely long time frame and a selection pressure to generate mutations that would stick to be able to produce a brand new molecule. It's simply not feasible to start from scratch. $\endgroup$
    – Joe Healey
    Jan 13, 2017 at 20:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .