Replicating the GCN eDNA test for other species: why it might not be the way forward

This probably seems like a strange topic for me to be writing about as founder of a specialist eDNA company, but one of the dangers for our nascent industry is that users don’t fully understand the limitations of the tools and throw the baby out with the bath water when they don’t work as expected.

Something we come across a lot at the moment is people asking whether a test like the GCN one can be developed for Species X. The answer is that yes, of course it can be. However, you should be aware that while a species test can appear to be whipped up almost instantaneously, it’s a huge amount of work to develop one that you can have real confidence in.

This is not to say that eDNA can’t help you (it probably can), but a test like the GCN one might not be the best option. I'm going to try to provide some technical background that can help you assess the best way to approach your project.

What sort of analysis is the GCN test?

The GCN test is a ‘real-time’ or ‘quantitative’ PCR analysis, known as qPCR. Crucially, qPCR doesn’t involve sequencing the DNA at all, it just detects whether or not amplification has occurred.

Hold on, what does all that mean?

PCR is the process by which we make millions of copies of - or 'amplify' a particular piece of DNA (in this context, a region of DNA that varies between species). It is one of the most common processes in molecular biology.

PCR uses primers, which are short segments of synthetic DNA that bind to either end of the region of DNA you are trying to amplify. You can think of them like book-ends; when they bind to DNA in a sample, the library in between them is copied. There are lots of good animations online that explain how this works - here's one example.

Primer design is fundamental to any assay.

The design of the primers is fundamental to any assay. We talk about primers a LOT. Depending on the question you are addressing, you can design primers to be highly specific to a particular species, or to be very general and match a much wider group. For instance, you can design primers to target ‘Atlantic Cod’, ‘bony fish’, or ‘animals’. Atlantic Cod DNA will be amplified by all three primer sets, Goldfish DNA by the second two, and Bumblebee only by the last one.

Primer design is initially done computationally, using existing DNA sequences of your target and closely related non-targets. Of course, how well this can be done depends on the availability of sequence data covering the right stretch of DNA both for your target and for related non-targets.

Following PCR, DNA sequencing can be used to reveal the series of bases (As, Ts, Cs and Gs) that characterise that stretch of DNA. For animals, we most often sequence a particular section of the Cytochrome Oxidase 1 gene (CO1, or Cox-1) which is 658 base pairs long. This essentially gives you a 658-letter word that describes your species.

When we sequence the DNA, it’s immediately obvious if something has gone wrong

When we sequence the DNA like this, it’s immediately obvious if something has gone wrong because the bases won't be made out clearly. Moreover, we can tell if we have accidentally amplified something that isn’t our target by comparing the sequence against reference databases. It’s surprising how common it is for researchers to accidentally sequence their own DNA instead of their specimen’s!

Beetle? Nope: fungus!

In our lab, we recently sequenced DNA extracted from an old, dry beetle specimen to try to generate a barcode. We knew we had successfully extracted DNA, and it amplified well (using quite general primers), but when we looked at the sequence it turned out that we had amplified fungus instead of beetle. It was only from analysing the sequence itself that we were able to know this.

So, what’s the problem with qPCR?

As mentioned above, qPCR doesn’t involve sequencing the DNA. Your question can be summarised as: ‘Is the DNA of Species X present in this sample?’ To answer this, you design very specific primers that only match Species X, and you run a PCR reaction. If Species X is present, the reaction will work and amplification will occur. Otherwise, if it is absent, no amplification will occur.

Because we never see the DNA sequence, we have to put a lot of faith in the primer design. We are assuming that the only species that could be amplified by the primers is our target (Species X), and we therefore infer from a successful amplification that the target is present. To be confident in this assumption, a lot of ground-truthing needs to be carried out on real samples where we have a priori knowledge about the presence or absence of the target, as was done for GCN by DEFRA and Freshwater Habitats Trust, culminating in the WC1067 report. Even then, we have to continue to be aware of this assumption we are making.

GCN are a fairly good target for a qPCR assay because they are well-studied, and there is therefore a lot of existing sequence information both for the GCN itself and for other closely-related species, which helps in designing highly-specific primers. Now consider that you want to target a particular species of water beetle (let’s call it WB1). You run into a whole series of problems:

(1)   Beetles probably shed much less DNA into the water than newts do, particularly as adults

(2)   There aren’t any available sequences for WB1 that can be used for designing your primers, so you have to find some fresh specimens and generate the reference sequences, using general insect or beetle primers.

(3)   You don’t know which parts of the DNA are particularly unique to this species, so you have to sequence several different genes in the hopes of finding a suitably unique region of the right length that you can design your primers around.

(4)   You have to do the same thing for all the closely-related species. There are a lot more closely-related beetles than there are newts.

(5)   It might happen that just by chance a completely unrelated species shares the same DNA sequence as WB1 (in fact, the GCN primers would amplify some African antelopes – we just don’t consider that too much of a risk in British ponds). This is very difficult to rule out, especially if the stretch of DNA you are using is not located in a commonly-sequenced gene for which there is a lot of available data.

(6)   There isn’t a lot of information about where WB1 is found in the natural environment, so ground-truthing is extremely difficult.

All in all, this is a huge amount of R&D work, which still might not result in a test that you can have total confidence in.

So, what’s the alternative?

Quite simply, sequence the DNA. Modern sequencing technology allows us to sequence the DNA of many different species in parallel, so it doesn’t matter if your primers are not totally specific. Sequencing will tell you whether you’ve amplified (a) your target, (b) a close relative of your target, or (c) several species that are all present in your sample.

the quality and quantity of information is far greater than from a qPCR test

The downside is that this approach takes longer and the costs are slightly higher, but the quality and quantity of information you gain is far greater than from a qPCR test, and there is relatively little R&D work required up-front.

If you'd like to know more, check out our website or send me a message.