Diabetes, the broad term for a handful of diseases that prevent the body from properly regulating blood sugar levels, was first documented over 3,500 years ago in ancient Egypt—yet experts still aren’t sure exactly how it develops, although scientists are almost certain that there’s no single trigger. Indeed, two primary forms of the condition are already known: types 1 and 2. Type 1 diabetes, which tends to have a more sudden onset, has proven particularly enigmatic, as people can develop the condition at different ages, and unlike type 2, it seems to be more closely linked to genetic and other predispositions than to diet and lifestyle.
Now, research published July 25 in PNAS may have revealed a key piece of the puzzle. The presence of the bacterium Parabacteroides distasonis in the gut microbiome causes type 1 diabetes in a mouse model and seems to predict the onset of the disease in humans. This is likely because the microbe produces a peptide similar enough to part of an insulin molecule that it can lead to the production of insulin-targeted antibodies, priming the immune system to launch an attack against insulin and the cells that produce it. Thus, the researchers have identified a microbial culprit for doctors to examine as they look for new ways to screen for and perhaps eventually prevent the disease.
Stanley Hazen, a Cleveland Clinic researcher who studies how the gut microbiome influences various diseases, applauds the study’s authors for going beyond merely identifying an association between a gut microbe and disease and actually probing the underlying mechanisms—adding that the common failure to do so makes many similar studies “junk.”
“Most investigations into the microbiome simply look at the types of microbes in the intestine or feces and show that the composition is associated with the prevalence of the disease,” Hazen says. “That’s just association, and . . . you cannot tell from that kind of analysis what’s the chicken and what’s the egg.”
Boston College biologist Emrah Altindis and his colleagues pieced together the behavior and functional role of P. distasonis bit by bit. It’s well established that the immune systems of people with type 1 diabetes attack insulin and the pancreatic cells that produce it. The team hypothesized that this autoimmune response may actually be an attempt to assail a foreign entity that’s structurally similar to insulin, which then goes awry. So they screened existing databases for sequences of peptides known to be produced by gut bacteria, keeping an eye out for structural similarities to insulin. After that screen identified over 50 candidates, Altindis explains, the team gradually narrowed the list based on the peptides’ degree of similarity to insulin and ability to activate insulin-attacking CD8+ T cells taken from a human patient with diabetes.
The team then moved to a mouse model, testing their short list of candidates by injecting mice with one of the peptides or with insulin and measuring their immune cells’ response. Out of all the possible peptides, only one, called hprt4-18 (which had already shown to be produced by P. distasonis), activated an immune response from CD8+ T cells in mice, Altindis says. The team then began another experiment in which they fed the bacterium to mice, seeding their gut microbiomes, in order to see how it affected disease progression. The specific mouse model used is fated to eventually develop type 1 diabetes, Hazen notes, but not as quickly as they did in this experiment. By the time they were 12 weeks old, mice colonized with P. distasonis showed clear signs of type 1 diabetes while controls, who were otherwise identical, did not. “We were able to accelerate the disease onset by just giving this vector,” Altindis says.
The goal is to help patients get treated, or to hopefully prevent new cases.
—Emrah Altindis, Boston College
Further investigation revealed that the newly diabetic mice had increased counts of CD8+ T cells and other immune cells implicated in type 1 diabetes such as dendritic cells and macrophages. Meanwhile, they had fewer of the CD4+ T cells that reduce inflammation. As Altindis phrased it, “the good cells are decreasing and the bad cells are increasing,” indicating that P. distasonis and its production of hprt4-18 had indeed triggered the kind of autoimmune attack that ultimately leads to type 1 diabetes.
“It is unclear what triggers the immune system to take that initial wrong turn,” University of Virginia diabetologist Heather Ferris, who like Hazen didn’t work on the study, tells The Scientist over email. “This microbiome-derived insulin-like peptide, which is close to self but not quite the same, could be that first trigger,” she says. “Once one antibody starts causing damage to the pancreas, other proteins that the immune system shouldn’t normally see are released from the pancreas and trigger more antibodies. So if you can stop the trigger then you could potentially stop the whole cascade from taking place.”
However, Hazen adds: “What this paper does not get at is ‘How big of a contribution does this mechanism [make] to type 1 diabetes in humans in general?’”
In a first step towards probing the human relevance of their findings, the researchers looked for the same trend in diabetes patients. They turned to the DIABIMMUNE project, a database that contains medical records for infants from Estonia, Finland, and Russia alongside demographic information and other potentially immune disease–relevant data, including sequencing results from microbiome samples taken at various ages. Among the 222 records examined, infants younger than three years old who had P. distasonis in their gut microbiomes had a greater risk of developing type 1 diabetes later in life (in the Russian and Estonian cohorts, 100 percent of the infants who were eventually diagnosed with type 1 diabetes had signs of P. distasonis in their gut), which Altindis says indicates that a person’s gut microbiome composition can serve as a powerful predictor of type 1 diabetes risk, though he stresses that the disease’s development is likely more complex, with other factors playing into it. The work, he adds, does not establish a causal link in humans, only the potential for one.
“We are never going to purposefully infect someone with this [bacterium] and see if they develop type 1 diabetes, so I am not sure that there is a definitive study to be done [to demonstrate causality],” says Ferris. “At this point, the mouse data is very good, but we have cured mice of type 1 diabetes 100 times over and it has never translated to humans.”
Still, “it is very exciting,” Ferris adds. One big caveat she notes is that the children in DIABIMMUNE are particularly homogenous from a genomics standpoint. “It will be interesting to see if this association holds up in more genetically diverse populations and in older patients, as the DIABIMMUNE cohort was 0 to 3 years of age and while this is a time interval during which many are diagnosed, the majority of patients are diagnosed after the age of 3.”
“I think the most important thing to make this data more convincing for its relevance to humans is replication in other patient cohorts,” Ferris says.
To that end, Altindis says that his team is analyzing other datasets to see if the association between P. distasonis and type 1 diabetes holds. “Then we will feel a bit more confident,” he says, though he notes that most of the data available for such analyses comes from the US and northern Europe.
Altindis, Hazen, and Ferris all say that it’s far too soon to talk about any therapeutic or clinical applications to the study, but that the research lays an important foundation for future work that may eventually reach that point, whether it be in the form of better screening for risk factors that may eventually lead to diabetes, uncovering novel treatments, or perhaps even a vaccine of sorts against P. distasonis that could be given to children genetically susceptible to the disease.
“Anything we can do to help people not have this difficult life. . . . The goal is to help patients get treated, or to hopefully prevent new cases,” Altindis says.