Overview: Biosynthesis of Neurotoxin Domoic Acid in a Bloom-Forming Diatom
In this post we continue our overview of the paper Biosynthesis of neurotoxin. Last time we learned that Brunson et al. (2018) had guessed that a redox enzyme was needed as part of the pathway Pseudo-nitzschia uses to synthesize domoic acid (DA). They then studied the Pseudo-nitzschia’s transcriptome to answer the questions: Which genes are upregulated under low phosphate conditions? and Of those genes, which are upregulated still more under increased pCO2 conditions? Since those are the conditions previously observed to induce DA production, genes fitting both descriptions were highly suspect to be involved in DA biosynthesis.
After locating CYP450, a redox enzyme, in this group, the authors located it in the genome to find it associated with others in a gene cluster. With four genes in a gene cluster now identified, three of which have known enzymatic activity, they could piece together a proposed biosynthetic pathway.
Like CYP450 (just “P450” in Fig. 1B), dioxegenase (the purple dabC) is another enzyme with oxidative reaction capabilities. The terpene cyclase (green dabA) is capable of N-prenylation, a kind of reaction that would react GPP and L-Glu to throw out the phosphate groups on GPP, take off one of the H atoms in the -NH2 group, and bind the two molecules together.
This information, on the chemical activity from these genes, led the authors to suggest this biosynthetic pathway. Let’s consult this diagram as we go through what they did before testing it to see if it was right.
Ok. Let’s break this down a little so we can follow the pathway. DabA does the first step, which just combines the first two molecules into a molecule called L-NGG.
With L-NGG made, we can move on to the oxidative reactions. See how dabC appears twice in the diagram? DabC can oxidize both L-NGG and 7′-carboxyl-L-NGG, although the reaction with 7′-carboxyl-L-NGG is more favorable a reaction. Nevertheless, if it reacts with L-NGG before dabD converts it to a new molecule, you will get dainic acid by-products. So this proposed pathway predicts that you’ll get a little bit of dainic acid on the side.
There’s one final step, however. Get the isodomoic acid and turn it into domoic acid. Isodomoic acid is an isomer of DA – that is, it has exactly the same set of atoms as DA, but they are arranged differently. Another example of a set of isomers would be glucose and fructose; both sugars, and both made of the atoms C6H12O6, but they taste different and are metabolized differently by the human body.
They suspected that dabB, the hypothetical protein heretofore unmentioned in the pathway, catalyzed the isomerization, but couldn’t find evidence of this kind of chemical activity from the protein. What is dabB and is it associated with DA production at all? What does convert the isodomoic acid to DA? These are questions asked near the end of the paper, and are a launching pad for future research.
But in the meantime, we have this entire biosynthetic pathway to get DA (well, isodomoic acid, but still, that’s pretty good) – and now we need to see if it actually works. Before continuing, note that a name like “dabA” can be used to mean two different things – when italicized, dabA refers to a gene; when printed in upright text, dabA refers to the enzyme that gene codes for.
The Test.
How did they test their new hypothesis?
The paper we have been studying was published in the journal Science. One of the really nice things about Science papers is that they are short, sweet, and to the point. Biosynthesis of neurotoxin is only three pages, and it’s packed with figures packed with high density, valuable information. The not-so-nice? You can’t describe most scientific experiments in three pages. You can’t present all the data you got in three pages and a handful of relevant, information-packed diagrams (as helpful as those diagrams are).
A Supplementary Materials document is often published along with the actual paper. It is less pretty, more verbose, and much longer (for this paper, the SM is ninety-two pages long!) But it holds the details to a lot of what is glossed over in the Science paper, and in our case, explains the methods used by the contributors.
Repeatability is very important for scientists. Everything a scientist publishes should have enough information for others to repeat what they did. So now, we are going to consult both the paper and its supplementary material to learn more about what they did in lab.
Transcriptome Sequencing
How do you sequence a transcriptome? How do you get the DNA sequences of genes that are turned on or off under certain conditions? The experimental set up started with cultures taken from Monterey Bay seawater. Pseudo-nitzschia multiseries was isolated as an algae out of microbes cultured from the sample, and then this culture could be used to inoculate new ones under various conditions.
The authors grew P. multiseries in normal conditions, limited phosphate, and increased pCO2 and extracted RNA from each culture. You can use special buffers to isolate RNA out of a cell culture while discarding the DNA. There are sequencing services labs employ to take and sequence their RNA samples. They will take the RNA, transcribe it into complementary DNA (cDNA) format, and sequence that. (RNA is very unstable, so the cDNA format is preferred. We also amplify the genetic material prior to sequencing, which requires a cDNA format instead of RNA).
This data is then given to bioinformaticians and data scientists who can process the data, identify the genes in use, and even quantify how much the genes are upregulated or downregulated in each sample. This allows for the creation of heat maps depicting upregulated genes. All 483 transcriptomes of upregulated genes under limited phosphate is on pg. 16’s Fig. S2 of the supplementary material, and of course narrowed down further to the few of these genes which were also upregulated under increased pCO2 in Fig. 1B.
Genes to Enzymes
After identifying dabA, dabC, and dabD and their potential biosynthetic pathway to DA based off of the transcriptome sequencing results, it was time to verify that these genes do, in fact, code for enzymes that can chemically convert the molecules GPP and L-Glu into domoic acid.
How did they do this? Well, there’s not much we can do with a gene. It’s a string of nucleotides. What we want is the enzyme the gene codes for. Those enzymes should be able to catalyze chemical reactions that result in DA production. So what’s really good at converting genes into enzymes?
Microbes, of course. Here the authors jump into biotechnology to get the job done. If you take the dabA gene and copy it out of P. multiseries’ genome, you can engineer it into E. coli and E. coli will produce the enzyme for you. Then you can extract dabA out of E. coli and run tests on the enzyme.
How do you get E. coli to pick up foreign DNA? It turns out that bacteria – many microbes, actually, including bacteria – have plenty of methods of exchanging DNA with their neighbors. One such way is transformation, and in it, bacteria exchange small circles of DNA called plasmids. The plasmids contain genes that benefit the bacteria, pressuring them to pick up and use this new bit of genetic information.
So the authors took a standard plasmid vector – a circle of DNA with a slot in it for inserting new genes – and copied dabA (and dabC) into the plasmid. This allowed E. coli to pick up the information and start expressing the genes for themselves. They then harvested the dabA and dabC enzymes from the engineered E. coli cultures and prepared them for experimentation.
There’s one other enzyme critical to the pathway, however: dabD. This enzyme, it turns out, is a transmembrane protein that only works when embedded in a membrane. E. coli can’t provide that. It won’t, therefore, express dabD well. How do you get around this problem?
E. coli is a popular organism used in labs; so is S. cerevisiae. S. cerevisiae is a type of yeast – so it’s a eukaryote, not a prokaryote; fungal, not bacterial. And it has microsomes, little vesicle-like pieces made up of endoplasmic reticulum. In other words, S. cerevisiae can carry enzymes in a membrane, while E. coli cannot. So dabD was engineered into that microorganism, and the microsomes were extracted so that the enzymes would still be functional.
Now you have enzymes (and enzymes within microsomes) in your laboratory. What do you do with them? You react them in a test tube – in vitro, if you’re feeling fancy – and run chemical tests on the results. It was here that they were able to confirm that dabA, dabC, and dabD do, in fact, react with substrates GPP and L-Glu to make domoic acid.
Chemical Tests
You made a chemical reaction happen. Yay! How do you know what’s in your vials?
LCMS chromatography is a means of separating out molecules based on how their interact with water. Chromatography separates out the components of a mixture. In LCMS, molecules move up a column atdifferent rates based off of the functional groups present in the molecule. The results for this test are shown in Fig. 3A from our paper. In this diagram, each peak represents a different molecule noted by a number.
- Domoic acid
- L-NGG
- 7′-carboxyl-L-NGG
- isodomoic acid
- (a,b,c) dainic acid
- 7′-hydroxyl-L-NGG
Each line, on the right vertical axis of the graph, represents different reaction conditions. Organic chemists on the team tried different mixtures to test the reactability of the enzymes with their substrates. So for example, the first two lines, a and b, are the tests run for dabA reactivity with the substrates GPP and L-Glu. In “a,” these two molecules and one enzyme were run with MgCl2; in “b,” only the molecules and enzyme were run. Thus in “b” there is a very small peak showing a little bit of L-NGG (consult the biosynthetic pathway as well as the LCMS graph to see what each enzyme is supposed to do; dabA is supposed to convert GPP and L-Glu to L-NGG).
The next reaction to test was the affects of dabC. Reacting dabC with 7′-carboxyl-L-NGG yields isodomoic acid when mixed with the chemicals αKG, FeSO4, and L-ascorbic acid (graph “d”), but when EDTA is added to the mix, it neutralizes the reaction and you only get the substrate (graph “e;” the substrate being the 7′-carboxyl-L-NGG). Throw the same three chemicals in with enzyme dabC but swap the molecule out with a different substrate – say, L-NGG – and you get dainic acid (there are three isomers of dainic acid produced here, all seen in the LCMS curve in line “c,” along with a small peak showing some of the substrate L-NGG left over).
Finally, the authors tested dabD by reacting the enzyme with L-NGG and a partner enzyme NADPH. Now, dabD was the transmembrane protein that only works if left in microsomes (graph “f”). So as a control, they also three empty microsomes in with L-NGG and NADPH (graph “g”). In “f,” you see a peak for the substrate, but also some 7′-hydroxy-L-NGG. To test that this actually was the molecule seen in the graph, they ran a chemical they knew was 7′-hydroxy-L-NGG to compare it (“h”). Now, they did get a very small blip in this graph that they’ve marked with a dotted line – a very small peak for the correct carboxyl group. So this molecule is getting made, but at very low levels. While they attempted to optimize reaction mixtures, it is likely the cell itself is better at making this reaction work than we are.
That being said, testing for abundance of four molecules (L-NGG, 7′-carboxyl-L-NGG, isodomoic acid, and dainic acid) in reaction mixtures shows a decrease of substrates L-NGG and 7′-carboxyl-L-NGG, and an increase in chemical products dainic acid and especially isodomoic acid, as shown in Fig. 3B. So this does, in fact, confirm the use of genes dabA, dabC, and dabD in DA production in Pseudo-nitzschia.
Conclusion
What did the authors discover? They found genes associated with DA production, and ran experiments after cloning them in E. coli and S. cerevisiae to confirm that they do, in fact, work in the DA pathway. As a double-check, the authors searched the genomes of eight different Pseudo-nitzschia species that have publicly available, mapped genomes. Only P. multiseries and P. australis have these genes, and those are the only two species among those eight which produce domoic acid.
There are questions that still need to be answered after this paper. Science is a lifelong pursuit, and papers often end with questions for other scientists to follow up on! But in this case, the authors went on to apply for – and win – a $5 million dollar grant to use to develop a new environmental monitoring tool, based off of this research. They were published in Science for their efforts, and now they are following up to apply their research to protecting human and environmental health.
This kind of work is exceptional in that it required a wide variety of expertise. Microbiologists (specializing in bacteria and in yeast), bioinformaticians, organic chemists, data analysts, and molecular biologists all needed to work together and collaborate to synthesize the information into a succinct experiment leading to such breakthrough results. This is a very interdisciplinary subject with big impact for human and environmental health, and it is by collaboration among experts in multiple fields that solutions can be reached and problems solved.
References
Brunson, J. K., McKinnie, S. M. K., Chekan, J. R., McCrow, J. P., Miles, Z. D., Bertrand, E. M., … Moore, B. S. (2018a). Biosynthesis of the neurotoxin domoic acid in a bloom-forming diatom. Science. https://doi.org/10.1126/science.aau0382
Bio by the Bay 
