Skip to main content

Drug Assays

Cryptic Natural Products Appearing

I last wrote about “cryptic natural products” here – this is the idea that there must be a great many interesting compounds produced by microorganisms that we have not seen yet. It’s for sure that there are many biosynthetic-looking gene clusters found in these species that don’t seem to be turned on most of the time, which makes one think that under the right conditions you could perhaps elicit some “break glass in case of emergency” structures that might be well worth seeing.

This was one of the original ideas behind the company Warp Drive Bio (here’s a presentation from 2016 on some of their work), but it’s still not one that everyone buys into. Philosophically, you wonder how many of these pathways are even activatable, and how many are just lost chunks of DNA that are still hanging around. There’s also plenty of debate about how, even if the idea is sound, one goes about getting the organisms to produce their hidden treasures. Here, though, is a new paper from two groups at Princeton that demonstrates the idea in practice.

They’re looking at Streptomyces, a genus that certainly produces some interesting compounds, and seems to have the potential to produce even more. The group used green fluorescent protein (GFP) as a genetic marker, inserting it into some of the “quiet” biosynthetic gene clusters as well as out in more neutral territory, and then ran all sorts of stress experiments on the organisms to see if any of these caused some activity. As it turns out, exposure to etoposide and to avermectin, both quite toxic to the organisms, caused some of these biosynthetic pathways to turn on, and several new compounds emerged, including one with antifungal activity and some that appear to be cysteine protease inhibitors.

What’s more, the group was able to then track down some new information about these pathways, showing that there’s a specific translational repressor protein that keeps them silent under ordinary conditions. That holds out some promise for being able to turn more of these clusters on, and under less drastic conditions. If there are other such repressors scattered around, and if they have any commonality, we could start to get a lot more control over expression of rare natural products. It’s still going to be a lot of work – one of the things that Warp Drive has reported on is that the presence of these gene clusters varies widely among different strains – but there might be a less brute-force way into the field. (I wonder, given that, if the molecules reported in this new paper have already been seen by Warp Drive’s internal efforts or not?)

Like any organic chemist, I admire (and am slightly terrified by) the diversity and complexity of the natural product world. Getting medicinally useful compounds out of it has become progressively more difficult as the years have gone on, in a perfect example of the “low hanging fruit” problem. If there’s a tricky way to cause new active leads to come spilling out of it again, I’m all for it. Not all of them are going to be useful, of course, and not all of them are going to do things that we particularly need done, but some of them may well be good things to have.

21 comments on “Cryptic Natural Products Appearing”

  1. Alex says:

    I’m a microbial ecologist PhD student working in this field. I think that it is very, very unlikely in my opinion that most of these biosynthetic clusters are “just lost chunks of DNA that are still hanging around” – while eukaryote genomes are messy and have plenty of pseudogenes, microbial genomes are usually quite streamlined and genes tend to serve a function (with the exception of symbiont and pathogenic microbes).

    1. Derek Lowe says:

      That’s a good point – I’m more used to eukaryotes!

    2. Dan Udwary says:

      Alex is absolutely correct. I can say that it’s fairly rare that natural product gene clusters are found broken, at least in Actinomycete bacteria. The operons/gene clusters encoding biosynthetic pathways are often huge and repetitive, and the bacteria almost certainly pay a large energetic cost in maintaining them. That said, they often aren’t expressed at levels high enough to detect products, likely because it requires even more energy to translate them and deal with potentially toxic products. So, I feel like it’s been known for a while that regulation manipulation is the key to making them productive, but it’s hard because (despite what Derek’s description of this paper might imply) there’s no magic regulatory gene that will work on every cluster, strain, or genus.

      1. SW says:

        Is there a reason you can’t try heterologous expression of these clusters?

        1. RQ says:

          I would expect these products wouldn’t be published in JACS if that method had worked. Seems like most of these easy solutions (heterologous expression) would have been published already if the product was seen that way.

        2. Fomer Merckoid says:

          I believe heterologous expression was tried with the Streptomyces avermectin system…

        3. Morten G says:

          Expressing one cluster doesn’t necessarily mean that you have all the proteins in a synthetic pathway since there could be crossover from other pathways.

          And figuring out which substrates to feed into the pathway (and/or individual enzymes) is near impossible, even with protein structures.

      2. Dan Udwary says:

        Genus-to-genus often works, but you can’t typically, say, drop an Actinomycete cluster into E.coli. Moving 100+kb without accidentally mutating it is not easy to begin with. The heterologous host often lacks self-resistance. There are post-translational modifications. Metabolic flux is going to be all wrong. And then there’s regulation. So, yeah, not impossible, but not achievable on a mass scale.

  2. yuriwho says:

    Perhaps, CRISPR to the rescue with an insertion of a unique restriction site to enable insertion of a strong promoter/enhancer in front of said mystery clusters?

  3. Some guy says:

    Just a random guy, but what about (instead of stressing the things), just trying to edit in a stack of promoters right before the genes of interest? Too high a risk of breaking everything? Doesn’t work like that in bacteria?

    1. Dan Udwary says:

      Sometimes you can! But in the worst case scenario, there are several dozen genes of interest in a biosynthetic cluster.

  4. gippgig says:

    Altering bacterial (and other) genomes requires 1 – getting foreign DNA into the cell & 2 – getting it to recombine with the genome. CRISPR helps with 2 but 1 still requires customized conditions for each organism. It’s like a finicky chemical reaction that requires optimizing reaction conditions for every particular set of reactants.

    1. H says:

      @yuriwho, Some guy & gippgig: CRISPR sounds quite powerful, and it also is if the machinery works. But it’s not a magical device that works in any bacterium you put it in. If you have a method to get CRISPR to work quickly in any organism, you can pick up your Nobel prize next year.

      1. Bagger Vance says:

        “When all you have is a hammer, every problem looks like a nail.”

  5. DanielT says:

    The way to do this is to screen for new pathways genetically, mutate the organisms with nasty mutagens, and then screen like crazy. Genetically screening removes the “we have seen this molecule 10,000 times before” problem and then you use brute force to get expression and screen.

  6. Toby Johnson says:

    I don’t understand why editing or integration into the genome is needed. Why couldn’t the cluster be combined with a constitutive promoter and then expressed in the cells, as a plasmid or just as free DNA. I imagine there might be a problem that the necessary precursors for whatever products the gene(s) might synthesize, might be present in nature but not in the culture medium. To me that seems a harder problem than “just” turning on an unexpressed gene cluster.

    1. Lora Cameron says:

      Depends on how large the cluster is and whether it needs any help assembling. Plasmids have a sort of maximum upper limit to how big a sequence you can fit in there before it doesn’t want to be taken up by the usual methods (cationic lipid, electroporation). You can break it up into multiple smaller sequences in individual plasmids, but then you have to do multiple selection, and while I’ve managed to do 4 selection reagents at once, more than that is a bit iffy. And then you have the trouble of keeping the plasmids all in there, they like to spit them out.

      The other thing is, sometimes the metabolites are produced only in response to specific stressors/signals: most soil bacteria behave differently in a biofilm environment than they would in pelagic culture (e.g. stirred tank). And there’s no great way to scale up biofilms in a controlled fashion. If you don’t know what is triggering the production of these things, well…good luck trying to find it, because the trigger may very well be coming from another species of bacteria entirely (e.g. luxS signaling).

  7. Natural Chemist says:

    I’m not convinced these silent pathways encode natural products that haven’t been seen already in other strains. I don’t know of many examples where heterologous expression or other genetic methods yielded something really unique. Can anyone help me with this?

  8. Insilicoconsulting says:

    waaaay back in 2001-2002 we isolated strains/species from samples in Western ghats, India, a known biodiversity hotspot. At the time we performed 16s rDNA analysis and many sequences were not similar to any existing gene sequences.

    Extracts (not single compounds) were assayed for various things and showed good activity in some cancer assays etc. Some of the compounds were isolated and continued to show activity at decent nM levels.

    Being more of an academic startup the money ran out and we could not continue this. But investors asked us exactly this. How are you going to coax out production?

    A common figure bandied about was that microbiologists had isolated less than 1 % of the total soil diversity so who knows what we would find!

  9. Former Natural Products Chemist says:

    I worked at Diversa (became Verenium, bought out by BP and BASF) in San Diego in the early 2000s and we modified a strain of Streptomyces for heterologous expression of biosynthetic pathways on cosmids. At the time we discussed all the issues above and tried to mitigate them as much as possible. To improve the host we disabled as many of endogenous pathways we knew of in the host strain. Interestingly when we put random cosmids in the host we activated many more pathways we were unaware (genome sequencing took longer back then). Also when we did the proof of concept experiments with known pathways the yields were really low. Internally we discussed precursor supply, correct/ strong regulation, lack of resistance genes etc. My view then was that this technology had to go back to academia for at least a decade. It’s been about 10 years and hopefully this is will help the technology move forward again. In general people are seeing >10 natural product pathways in the Actinomyces/ Streptomyeces species that are sequenced.

  10. Dr. Manhattan says:

    Papers from Sean Brady and others describe the use of TAR Transformation Associated Recombination (in yeast) to reassemble soil bacteria NP gene pathways. Very much faster and high throughput than traditional cloning techniques. The use of TAR also allows these pathways to be placed directly in E. coli/Streptomyces shuttle plasmids.

    See: Montiel et al. Proc Natl Acad Sci U S A. 2015 Jul 21;112(29):8953-8958 and Katz et al. J Ind Microbiol Biotechnol. 2016 Mar;43(2-3):129-141.

Comments are closed.