This is part 2 of an interview with Aleksandar Kostic, a postdoctoral fellow at the Broad Institute of MIT and Harvard, who uses powerful computational approaches to study microbial populations. You can read part 1 here.
What potential clinical applications are there for this work? Either before children get type 1 or maybe after they get it?
The way I look at it is, this is an important study because really it’s the first to show a defined effect between the development of the microbiome in infancy and the diagnosis of type 1 diabetes. What’s important about it is that it opens doors to further questions. What I think will come out of this are targeted, prophylactic treatments rather than something that’s meant to cure disease. Scientists need to dig further into what are the important species that are protective in these children. How do they relate to the bacteria that are protective in the NOD mouse model? How and when can we transfer the good bacteria and prevent people from developing the disease? These are questions that people are now addressing in other diseases— Clostridium difficile is the major one. Right now, the treatment for this is usually ingesting an intestinal slurry, but this is such a serious disease, and the effects are remarkable, so people will do the treatment. It’s a 95% cure rate of a disease that normally doesn’t go away and kills people. So I think that’s an important disease in opening the window in terms of how you use communities of microbes as therapies. As this field develops, the probiotic organisms will be delivered in tablet form rather than in a slurry.
So you’re saying that once you know which bacteria are protective, you could maybe have a pill that people would take, before they have seroconverted?
That’s something that’s conceivable, even for children who are just at risk for developing the disease, genetically. If it’s a safe therapy, why not give it prophylactically just like you would get a vaccine?
And what about post-diagnosis? Is there any thought this might help with that?
Post diagnosis in type 1 diabetes is much more difficult because at that point the beta cell destruction has already begun. So that’s an important question, and I think unfortunately right now, it’s outside of the realm of what the microbiome can deliver on. It’s outside of my field, but I do see a lot of promising things happening in the stem cell field and the real possibility of rebuilding beta cells using stem cells from patients. But again, I think that addressing patients with disease is going to be a much more difficult challenge.
Is there thinking that the microbiome is related to the autoimmune attack that’s part of type 1?
From what’s currently understood about how the microbiome interfaces with the immune system, once the beta cell destruction has begun there’s little that can be done at that point, because the destruction is being mediated by t-cells.
So it’s kind of a different issue in the body.
Yes. Not that it’s something that I think will remain unaddressable, it’s just that right now, where the microbiome field sits, because it is so new, really the only interactions that are well described are the microbiome with the innate immune system, meaning regular epithelial cells, macrophages, dendritic cells, but not t-cells. It’s certainly possible to think a few years down the line, as the interaction between the microbiome and the immune system is better characterized, that these kinds of things are possible, but it’s not in the near future.
From what I’ve read the average age of type 1 diabetes diagnosis has been dropping. Does this work explain that at all or give any hints about why it might be happening?
It might. And it feeds into our next study that we’re finishing and submitting this week for publication. The study is related, but it’s much bigger. It ties into the hygiene hypothesis. We have stool collected from children from birth to 3 years old—74 children from Finland; 74 from Russian Karelia, right across the border from Finland; and 74 from Estonia, which is nearby. What’s really interesting about this cohort, especially about the children from Finland and Russian Karelia, is that they share a lot of similarities—genetically, socially, they even speak the same language. But there’s a huge difference in their incidence of autoimmune disease and allergies in general and type 1 diabetes in particular: the rate is eight times higher in Finland. And the thought is that it’s simply a matter of their lifestyle; the Russian Karelians live a more traditional lifestyle, where they for the most part live on farms and are exposed to farm animals and soil bacteria, and non-processed foods and less refrigeration, a more natural upbringing. All of the Finnish children are urban and have a higher socioeconomic lifestyle. We see huge differences in the microbiome between these two populations, and we believe the differences are strongly tied to the differences in hygienic environment between the two places. In other words, we see a very big difference in the kinds of bacteria that are seeded in an early age in the Finnish children that are essentially absent from the Russian Karelian children. We think these bacteria lead to autoimmunity and the allergic responses that are seen in the Finnish children and in most Western countries.
This ties into your question of why people are getting it earlier and earlier. One of the most striking figures I’ve seen is if you look at the incidence of infant infections, very severe life threatening infections that happen in infancy and you look at where these things happen in the world on the map, as you’d expect they’re completely absent in the West and fairly prevalent in Asia, Africa, and South America. But then if you superimpose a map of type 1 diabetes in the world, it’s the mirror opposite. So you kind of take the good with the bad. I think the early bacterial colonization of those children who live in those environments predisposes them to severe infections, but I think it is an essential part of their immune development that later on prevents their immune system from attacking their own body.
So maybe the immune system’s getting an early challenge. Does the same thing hold true for asthma, allergies, and other autoimmune conditions?
Yes, all the hypotheses tie together. In general autoimmune disease and allergic disease are lumped in one category, and the basic mechanisms that set them off are probably pretty similar. Type 1 diabetes might be the strongest correlation, but the map is going to be very similar with other autoimmune and allergic diseases.
How do you actually study this? What is the process for understanding the different bacterial communities?
It’s all based on DNA sequencing. What we do is take the bacterial communities, remove the DNA, then sequence pieces of the DNA by what’s called random shotgun sequencing: you break it up into random pieces and sequence it. When you do this you have these little samples of what each community is made of. What has really fueled this field is technology that allows you quickly to read these pieces of DNA and tell which bacteria species they’re from.
How do you read them?
There’s a machine that takes all this DNA, stands it on its head, then reads out each base one at a time, ATCG. It reads it out in huge multiplex, hundreds of thousands of DNA strands all at once. That’s what allows us to look at this whole microbial community in very high detail and see exactly what’s there.
How much data are we talking about?
One stool sample from one patient, the amount of sequence data we get out of that is one terabyte of data. An iPhone holds about 16 gigabytes, so this is about a hundred times that. Just from one sample. So what that means is the analysis is very computationally intensive. We have a farm of servers that do all the number crunching and all of the comparisons between people.
It’s a huge amount of data; it is very cheap to sequence, but it’s too expensive to store anywhere and too expensive to analyze. That’s where the field of genomics is right now: we have so much data we can’t even process it.
I was wondering about the timing of the microbiome field, when you were starting out doing your PhD, was that right when people were getting excited about the microbiome?
Yes, the microbiome, it’s weird, I think of it as a new field but really it’s a renaissance. It used to be very popular in the fifties and sixties; people thought about it a lot back then, and would try to grow the bacteria cultures to study them. Then it kind of disappeared. And now, with DNA sequencing, for a given person we can tell you exactly which bacteria species you have, and that’s never before been possible. High-throughput sequencing started to become financially doable in 2009, and that was really when people started to become interested in the microbiome again.
You wrote in your paper that there were “significant shifts that occurred within T1D cases including an increase in the multiple sugar transport system.” My ears perked up, because I wondered if these changes could affect how someone processes sugar?
We don’t understand it very well yet. It just happened to be that the strongest difference we saw was in how the bacteria transported sugar. The amount of free glucose and other sugars in diabetics and maybe even prediabetics is very different than in healthy individuals, so it’s possible that the microbial community is different simply for that reason.
Could the microbial community be creating a difference in how people process sugar? I know that with my daughter, once she’d been diagnosed, we realized looking back that her desire for sugar had really changed, clearly something was really changing in her months before she was diagnosed. She wanted more maple syrup; things that used to be sweet enough for her, like berries, no longer were. It was like she wasn’t processing the sugar as efficiently as you or I would.
That’s really interesting. That could very well tie into the microbiome. There’s a lot of new research on the microbiome and its potential connection to the brain, and there’s a theory that the microbiome influences what you decide to eat. So for example, it’s easy for me to imagine that a microbial community that thrives on saturated fats because it’s their preferred food source has evolved a way to communicate to the human brain, Hey look at that juicy steak, I want to eat that. I really do believe that the microbiome can influence your choices of what to eat; I’ve seen it myself personally. For various crazy reasons, about six months ago I decided to go low carb, and I now have about 20 carb a day—no sugar, no grains, very little fruit. What I noticed when I first went on that diet was that it was incredibly hard in the beginning, and I was having dreams about cupcakes and all these kinds of foods. I really feel like it was my microbiome influencing it, because three weeks later those cravings went away. I think it could be that a certain constituent of my microbiome died out at that point, after it stopped being fed all those carbohydrates, and it’s no longer around to signal to my brain that I have to have that loaf of bread. It’s a theory, but it’s one possibility.