Type 1 Diabetes (T1D) is characterized by autoimmune destruction of the pancreatic beta cells ordinarily responsible for insulin production. 70% of people who get T1D have risk alleles, but only 3% of people with those risk alleles get T1D, suggesting that environmental triggers play a major role. One prevailing theory is that some pathogens produce proteins which resemble those expressed by pancreatic beta cells, duping the immune system into engaging in "friendly fire" if a susceptible person is exposed to the pathogen at the right time in development.
By scanning bacterial genome databases for proteins which resemble two of the major T-cell autoantigens in T1D (GAD65 and IA2) I discovered that Legionella—a known bacterial pathogen—expresses a protein which fits the bill. It's not a perfect match, but close enough to warrant further investigation. Legionella is a common contaminant of certain kinds of air conditioning systems, as well as potable water lines. It's best known for causing the rare lung infection Legionnaire's disease in the elderly, but exposure to it is surprisingly common; the bacterium can thrive anywhere warm water stagnates, so it's found in hot tubs, ice machines, showerheads, water lines, fountains, etc.
Subclinical Legionella exposure offers a compelling explanation for the peculiar epidemiology of T1D, including seasonal variation in diagnosis, the bimodal latitude distribution of high-incidence countries, and the neighborhood-to-neighborhood variation in incidence rates within cities that defies SES patterns typical of e.g. Type 2 diabetes.
For the full hypothesis and more details on the preliminary evidence, see this piece.
Exposure to Legionella only causes full-blown legionnaire's if you're severely immunocompromised, but in most people it produces a transient flu-like illness known as Pontiac fever. I'd like to investigate the possibility that this is the trigger for T1D in susceptible people.
I am a microbiome scientist and not an immunologist, but this year finds me in a visiting scientist position at a world-class immunology institute in Switzerland, so it seems like the perfect time to leverage the resources and expertise available to me to do some high-potential-impact research.
[Minimum funding]:
Experiment 1: Anti-Legionella Immune Reactivity in Patients vs. Controls To evaluate the potential involvement of Legionella in the pathogenesis of T1D, we'll start by recruiting newly diagnosed T1D patients as well as age-matched healthy controls, and obtain blood, sputum, and fecal samples from them. We'll get a panel of diverse Legionella species from a public bacterial strain bank, grow them in culture, and test the blood of patients and controls to see if it contains antibodies that react with any of these Legionella species. If patients are immune-reactive at a higher rate than healthy controls, it tells us that they've been recently exposed.
Experiments 2 & 3: Presence of Legionella DNA in Patient & Control Samples With the fecal samples and sputum, we'll run a PCR test that will give a highly specific and sensitive "yes/no" answer about whether there's Legionella DNA in the sample. If this is the case more often in patients vs. controls, that would further support this hypothesis, and give clues as to likely routes of exposure.
---
While not conclusive, positive results in any of these experiments would be enough to give this theory some momentum. Hopefully, this will spark related investigations and replication attempts by other labs, and start the process of moving us toward a consensus understanding of T1D's etiology.
[With goal funding]:
Experiment 4: Serogroups Even bacteria of the same species can have substantial differences, and these differences can impact diagnostic accuracy: existing urine tests for Legionella pneumophila exposure only work well for one "serogroup". If experiments 2 or 3 (stool & sputum PCR) reveal higher prevalence of Legionella in patients than in healthy controls, we'll plate those samples on selective agar to isolate the bacterium, enabling us to explore the interaction between patients' immune systems and the exact strain they're fighting. Whole-genome sequencing of these bacteria may help identify strains or features more likely to cause T1D. If PCR is positive but live organisms can't be recovered from patient samples, we'll travel to sample their homes following CDC guidelines.
Experiment 5: Demonstrating Cross-Reactivity Antibodies and T cells are generally very specific for their targets. If the results of experiments 1 - 3 support Legionella involvement, we'll tag and identify the specific components of the microbe that the immune system reacts to, and purify the immune factors that recognize them. Then, we'll be able to test if the immune factors that stick to Legionella are the same ones that react to the human proteins targeted in T1D. If so, this would be pretty conclusive proof of the hypothesis, and I'll thank you all from the podium in Stockholm once I'm close enough to death to get a Nobel.
All assays will include negative controls, which will involve testing unrelated bacteria and non-T1D-targeted human proteins alongside the subjects of this hypothesis.
[Beyond Goal]
Experiment 6+: Remediation, Prevention Islet cell antibodies can emerge months before the clinical onset of T1D, suggesting that chronic exposure to the trigger may play a role—and that it may be possible to intervene in this prodromal window. Chronic Legionella exposure is common in a variety of settings, such as when a building's evaporative air conditioner is poorly maintained, or a water heater fails to reach kill-temperatures in distal portions of the water line. Transient exposure may also result in colonization of the gastrointestinal tract or other organ systems, creating an internal source of chronic exposure. If earlier experiments are indicative that Legionella is a driver of T1D, we'll work with patient advocacy groups to recruit a cohort of individuals who show prodromal signs of the disease, identify sources of Legionella in their local environment, and implement control measures to eliminate the contamination where possible. These individuals, and a control cohort of equal size, will be tracked longitudinally to determine conversion to T1D status. If a significantly lower conversion rate is observed in the intervention cohort, this would suggest an eventual path to eradication of the disease, by informing the design of future building code and maintenance policies, design of equipment such as air conditioners, and greater vigilance by the general public around other known sources of Legionella exposure such as hot tubs.
Funds will be used to:
Obtain a panel of reference strains of common Legionella species
Purchase reagents for assays, synthetic peptides, & materials for sample collection
Cover travel costs involved in sample collection
Sequence genomes of patient-derived Legionella isolates
Pay the goddamned piper for open access publication in a high-impact journal
The most similar project I've got under my belt is a large-scale effort to identify anti-inflammatory and analgesic bacteria from the human gut microbiome.
I conceived of the program, designed it (from in vitro assays through molecular/mechanistic characterization and on to animal studies), wrote the grant, and landed my company $1.5M in funding to carry it out. I then worked with the rest of our team to grow 250 strains of anaerobic bacteria and process them into a format amenable to high-throughput screening by my colleagues. Hits from the primary assays did quite well in the animal trials, and—last I heard—are being shopped around to potential licensors in pharma.
All this was at my last company, but hopefully it demonstrates that I can create a functioning experimental pipeline and slot into it where I'm most useful.
Currently, I'm at the Swiss Institute for Allergy & Asthma Research, (SIAF) headed by Prof. Cezmi Akdis—a former pediatrician, world-renowned immunologist, and expert on T-cell-mediated immunity who is advising on this project's experimental design. The institute's staff consists of some three dozen scientists whose expertise collectively covers every domain of molecular immunology. While the whole institute is a very collaborative environment, I'll be assisted in this effort primarily by my colleagues Dr. Juan Felipe López Crespo (MD, MSc, PhDc) and Dr. Yagiz Pat (MD). Dr. López Crespo has extensive experience in molecular immunology methods and conducting human subjects research, while Dr. Pat brings expertise in clinical microbiology.
Hypothesis not true:
This wouldn't be a real failure, since the goal of the project is to investigate the hypothesis, and if it turns out not to be true, great—we can cross it off the list of possibilities.
Most likely causes of actual failures:
High background rate of immune reactivity against Legionella (or high prevalence in samples from healthy controls) makes it impossible to conclude anything without unreasonably large sample size. PCR & culture are highly specific, so there's little risk of false positives.
Strain variation in immune recognition yields false negatives in experiments 1, clearance of Legionella by the time of sampling yields false negatives in experiments 2 & 3.
Outcome of 1 would mostly be a need for more funding, or potentially a redesign of the methods before this can be investigated properly. 2 is more serious, as we'll try to publish even if findings are negative. A false negative could steer people away from investigating this possibility for years to come. This seems like a small risk, given nobody's investigated it so far anyway.
None.
(Help me Obi-wan Kenobi, you're my only hope.)