This is insane. This is either one HELL of a coincidence, or we have a team of researchers from Boston College and others from Harvard, the University of Florida, Seattle, and Auckland, NZ who are really on to something here.
I just got tipped off to this yesterday in the weekly literature review in Science, but the original article came out late this summer in the Proceedings of the National Academy of Sciences while some of us were apparently too busy sipping fruity vodka drinks out on the deck. I got a reprint yesterday afternoon about 5 minutes after I asked for it from corresponding author Dr. Emrah Altindis at BC, so I’d like to thank him for that.
It seems that a common bacterium found in the human gut makes a protein that looks just enough like insulin to cause the immune system to attack not only this bacterium, but indeed to attack insulin and the pancreas itself. This process could very well be one of the main triggers of Type 1 diabetes, and knowing this would substantially change the way we think about this disease. And it all may be just a crazy accident of nature.
The pancreas is dotted with “islets of Langerhans”, or simply “islets”. Most of the cells within each islet are insulin-producing β (beta) cells. In Type 1 diabetes, T cells mistakenly attack these β cells and destroy them, so we can’t make enough insulin, and we lose control of our blood glucose levels.
That most often happens because antibodies against insulin develop in the body. But why does that happen in some people and not others? What triggers it? We don’t really know.
Rates of Type 1 diabetes have been rising faster than genetics can account for, and the concurrence of Type 1 diabetes between identical twins is only somewhere between 30% and 70%. Another interesting anecdote is that in the Karelia region straddling Finland and Russia, Type 1 diabetes occurs about six times more frequently on the Finnish side than the Russian, despite similar genetic makeup in the two regions. So while genetics may predispose someone to developing the disease to some degree, the assertion that “it’s genetics” just isn’t a satisfying explanation.
Insulin is made up of two protein chains called “A” (blue below) and “B” (green):
The green spiral part — you’ll see it referred to in medical circles as “insB:9-23” — is by far most commonly what antibodies in patients with Type 1 diabetes are directed against (more than 90% in diabetic mouse studies, anyway). Is that because this part of insulin is simply most recognizable or accessible? But lots of other proteins in the body are recognizable and accessible, too, and they don’t get attacked so frequently. So, what’s going on here?
In 2018, a study published in Nature took a look at gut bacteria that were most commonly overrepresented in children that had developed Type 1 diabetes compared to normal gut bacteria profiles. The bacterial genus that led all others in that respect (P < 0.001, or 99.9% likelihood of a real effect) is called Parabacteroides. Not clear why that would be, exactly, but keep that name in mind.
In this study, Altindis and colleagues set out to find any proteins among all the known viruses, fungi, and bacteria that might look enough like insulin to trigger an immune response. (We have so much genomic information these days that doing this is not hard at all anymore.) This was a total fishing expedition; they weren’t biased for or against anything in particular. Of all the zillions of known protein sequences out there, they found just 47 candidates with stretches that resembled insB:9-23 closely. Then they narrowed that list down to 17 especially good ones, all from different species, where a lot of the amino acids matched the most important ones known to attract antibodies to insB:9-23.
So they manufactured these 17 peptides (protein snippets of length 15, the same length as insB:9-23) and exposed each of them to human T cells that were already directed against insB:9-23. They also included insB:9-23 itself as a positive control and a random peptide as a negative control. Of course the insB:9-23 peptide evoked a response from the T cells. But only one other peptide out of the other 18 did: the one from a well-known gut bacterium called ... Parabacteroides distasonis 33D.
Parabacteroides, the very same type of bacteria with the strongest correlation to Type 1 diabetes development in children.
Holy cow. No way. Talk about finding a needle in a haystack.
The Parabacteroides peptide (which the authors now call hprt4-18) resides within a protein that actually has nothing to do with insulin at all. That protein, hypoxanthine phosphoribosyltransferase, allows these bacteria to salvage some waste molecules to make RNA and DNA out of. Plenty of other bacteria have this enzyme, too, but their versions don’t have the insulin-like snippet. Not even all species of Parabacteroides have it. The whole thing seems like an innocent mistake. Maybe Parabacteroides, living in our guts for millions of years, borrowed this snippet from us and stuck it into one of their proteins and the whole thing worked out nicely … for them, anyway. I’m sure that now some evolutionary biologists will try to figure out how that came about.
Here are the two snippets — ours and theirs — side by side. I got help drawing out the structures from Tulane University’s online PepDraw feature:
Because the problematic sequence hprt4-18 is now known, the authors were able to go back through a diabetes database that contains both clinical outcomes in children and genomic sequence data from their stool samples and look for the hprt4-18 sequence specifically. And wouldn’t you know it, the kids whose stool samples harbored the hprt4-18 sequence were far more likely to develop Type 1 diabetes:
This appears to be the first time a direct link has been shown between the onset of an autoimmune disorder and a sequence from another organism that mimics a human sequence.
OK, where can we go from here? I’ll leave that to author Dr. C. Ronald Kahn, also a member of the BC research team:
“A plethora of human gut microbiome studies have demonstrated that the composition of gut microbiota in patients with autoimmune diseases, including multiple sclerosis, inflammatory bowel disease, and others, are significantly different from those in healthy controls. Our findings demonstrate a new link in which there is molecular mimicry between a peptide made by normal gut microbes and the autoimmune response in T1D. This suggests the potential to develop new tools, including vaccines, antibiotics, or probiotics, for the prevention and treatment of T1D and perhaps other autoimmune diseases.”
I know there’s a ways to go before this translates to any therapeutic success, but the best place to start is knowing the culprit. Here we have a common gut bacterium that is strongly associated with the development of Type 1 diabetes, and it turns out this particular bacterium — of all the bacteria, fungi, and viruses out there — just so happens to have the one sequence that looks enough like insulin to elicit an immune response from T cells directed against insulin. Of all the gin joints in all the towns in all the world…..
The smoking-gun nature of it is astonishing, don’t you think?