A conversation at the forefront of human microbiome research

What kind of changes have you seen in microbiology, particularly in the human microbiology fields?

It has changed a lot since I first started in 2002. Back then it wasn’t really such a big and busy field. There’d been lots of nice work in the microbiota for well over 100 years, but very small scale relative to now. When I started, there was maybe one microbiome paper every month if you were lucky. It might be a rumen paper or something, but you’d be quite happy to see any sort of paper on a host-associated microbiome. Now, and since the advent of next-generation sequencing, it’s gotten crazier and crazier. Last year, I think there were around 14,000 gut microbiome papers, so over 1000 every month.

How far away are we from fundamental studies to developing microbiota-based treatments to target Inflammatory bowel disease (IBD) and other such diseases?

One of the major drivers of current research around the world is the question, how do we harness the microbiota for therapies? There’s a lot of promise, but also a lot of hype in that area. Many spin-out companies have been formed to address this problem too. Alas, some of those companies have gone out of business because their ideas were not as promising, or they weren’t able to deliver as fast, as originally hoped. There are some notable successes though. For example, I am now in my mid-40s, and I think I can likely cross off C. difficile disease as a potential cause of my own death! By the time I’m old enough to have to worry about C. difficile, I think that will be a solved problem. There are already licenced treatments available, and more will come.

Beyond that, I don’t know. IBD is a very difficult disease. I think with C. difficile, it’s a relatively easy problem to solve because you have the pathogen, you get rid of the pathogen, and the disease goes away. With IBD, there’s underlying host pathology in there as well. It isn’t entirely about the microbiota. Furthermore, because every human being has their own unique microbiota, it’s very difficult to find treatments that will work for everyone and will work all the time. I’m fairly sure over the next 20 years (which, by the way I think is more realistic timescale than five years), we will have quite a few more effective microbiome treatments. But I think along the way, lots more companies will fall by the wayside because their products didn’t stand up to testing or clinical trials. This is of course completely normal in drug development, and the best treatments will hopefully rise above and make it to the clinic.

When discussing the microbiome and health, it’s really complicated science because everyone’s microbiome is different, and it makes it very difficult to find consistent stories. With this in mind, my usual advice to people starting out is to be wary of people who talk of the microbiota in definitive terms, and give simple soundbites. The truth is very rarely that simple, and there is a huge need for nuance when discussing the microbiome.

Adding to that, I wonder whether precision medicine will be the future - to have more targeted treatments for each IBD patient.

In an ideal world, that’s great. Everyone goes to the doctor, they get their microbiome profiled, they get their genome sequenced, their metabolites profiled, and they get a treatment that is tailored specifically for them. A major problem with this though is that it is currently too expensive for widespread implementation at a population scale.

The second problem is that it makes things more difficult for industry. If you do a drug trial, you have to show a better outcome than with a placebo. This is trickier if only a small proportion of people show a beneficial response. Likewise, if you develop a therapy that only works in 5% of people, that is a much smaller potential market, and subsequently, a much smaller payoff for the developers, unless the product is priced at a much higher level. Now it’s being done with things like immunotherapies and cancer, but I do worry that these fundamental problems will discourage some innovation in the space because companies don’t see a profit in it.

The third barrier is you need to have clinical practitioners who understand the microbiome. You cannot necessarily expect someone working in a routine hospital diagnostic lab to be an expert in metagenomic profiling, and understand what the resulting profiles mean. So, while I think that precision medicine has a huge promise. There are still significant barriers to its implementation. I definitely see potential in every patient getting their microbiome sequenced, but I don’t see that being routinely included in treatment decisions at a population scale for some time.

What are some of the outstanding challenges to conducting studies in the microbiome field?

I think that lack of reproducibility is a key problem for microbiome research. Lots of correlative findings, including landmark discoveries like links between the microbiome and obesity, have not reproduced very well. Often, this can be attributed to the problem of inter-individual variation in microbiota composition between people. Small-scale, underpowered, studies are sometimes not particularly helpful (and I say this as someone who has done plenty of this sort of work myself in the past). They might set you down a path of investigation, but sometimes that turns out to be the wrong path.

I think there are a couple of main potential solutions to this problem. The first is that we likely need bigger studies. Ideally, this would include hundreds or thousands of participants, and be verified independently across multiple populations. However, this takes a huge amount of time and funding, and so is often not practical. The more pragmatic solution for me is to follow up on correlations derived from smaller-scale studies by doing mechanistic experiments in the laboratory.

How does your lab study the functions of the microbes?

We are interested in the functionality of the gut microbiota. We want to know what individual gut microbes are doing, how that depends on context, and how the microbes interact with each other. Ultimately, we want to better understand some of these dominant obligate gut anaerobe species, many of which we actually know very little about currently.

In my lab, we do the top-down sequencing approaches to generate hypotheses. We also often then go back into the lab, grow the bugs, do experiments with them, and test hypotheses that come from sequencing. To my mind, this combinatorial approach offers the best way to truly get towards a mechanistic understanding of the gut microbiome. We are lucky that, at the Rowett Institute, over a period of decades we have built up a large culture collection containing thousands of different gut microbiota strains, and this often allows further study of the actual bugs. This has allowed us to go all the way from human studies, where we’ve done sequencing to find gut bacterial species that are associated with particular diets, through to working with the bugs, verifying that they do indeed use that dietary compound, and eventually going right down to identifying the enzymes that those bacteria use to break down the dietary compound. When you put that all together, you’ve got a very strong evidence base that these bacteria are carrying out a particular function in the gut.

With this in mind, when giving career advice to early career investigators, my plea is always to do your best to broaden your skill set, and not to just rely on sequencing. Sequencing is great, but for truly powerful mechanistic understanding, combination with other techniques is even better. It is true that gut anaerobes can sometimes be difficult to grow, and you do need a lot of specialist equipment to do that. However, if you can’t grow the bugs, you can perhaps think about other approaches, such as visualisation techniques, protein assays, etc, anything to add to the sequencing and help you address mechanisms. Sequencing alone may only get you so far.

Computational and bioinformatics tools significantly advance our studies. What are your opinions about the omics approach to addressing “correlation and causation” as you discussed in your talk? 

The key point is that both computational biology and wet lab work are important, and should be complementary! The main consideration for me though is always that correlations are not necessarily meaningful unless you can back them up with mechanisms. Omics approaches are super powerful, but they are often more hypothesis-generating than hypothesis-proving tools. If you find metabolite X is associated with a genome or assembled genome Y, that doesn’t necessarily prove that species Y is making the metabolite. Ideally, if you can get that bug from a culture collection, you can run tests to see if it actually makes the metabolite. Quite often you find that it doesn’t, the original omics-based finding was just a random correlation, or a mistaken annotation. To be clear, there’s absolutely nothing wrong with data correlation. It is a logical use of these hugely powerful techniques. But like I say, it may only get you so far. Ideally, follow-up verification work can be done to prove that what you’ve seen is actually meaningful.

What does the future of the field of human microbiome studies hold? Do you have any concerns?

In the short term, there’ll be lots more sequencing. They are amazing technologies, and people are still keen to fully exploit them. Being honest though, I do worry a little about the future of the field, because I fear we are now creating a situation that selects for people who are really good at sequencing/bioinformatics, potentially at the expense of those who do lab-based biology. To give an illustrative example, you could either spend your entire PhD classifying one protein, working on its function, finding the structure, or you could spend it working on tens of thousands of metagenomes that are publicly available. Chances are, the metagenomes work will end up in a high-impact journal, while the lab-based functional work will end up in a far lower-impact niche journal.

Then, when it comes to things like subsequent fellowship applications, who is more likely to get funding that sets them on the path to becoming an independent Principal Investigator (PI)? The metagenomics person with multiple Nature/Science/Cell papers, or the lab-based person who’s got one paper in a niche journal? Are we going to create a situation where we have lots of PIs who are great with sequence data, but do not have much lab experience? As I alluded to earlier, science really moves forward as a result of collaboration and complementary skill sets. Sequencing will only get you so far, and results are only ever as good as the reference databases we can map the data back to. Those reference databases fundamentally rely on people doing experiments in labs, and we also need people to properly verify sequence-based results. I do have concerns that, as people with these skills retire, they will not be fully replaced. I don’t think this would be a good thing in the long run for microbiome science.

Nonetheless, on the more positive side, I do think more and more people are acknowledging that sequencing will only get you so far. More groups these days are doing complementary lab-based experiments to support their sequencing work. So, my advice to early career researchers is to try and do both if you can. I do acknowledge though that this can be difficult, and can often depend on the available resources and expertise within individual labs.