Healthcare

IBM and Johns Hopkins University School of Medicine Discover Unique, Pathogenic Autoimmune Cells in Type 1 Diabetes

Share this post:

Computer simulation showing the extremely tight binding to an immune system cell of a newly discovered peptide (blue), which could cause the mistaken destruction of healthy insulin-producing cells in the pancreas, and in turn, type 1 diabetes. This autoimmune response is 10 times stronger than that seen with a weaker-lab-engineered insulin mimic (red), that is itself 1,000 times more immune stimulating than normal insulin. Credit: IBM Thomas J. Watson Research Center

Type 1 diabetes (T1D) is an autoimmune disorder where the body’s own immune system destroys insulin-producing cells in the pancreas. Currently, there is no cure for T1D, as the underlying interactions responsible for the disorder remain unknown.

However, the rise of immunotherapy research is helping us to better understand the immune reactions that cause T1D and holds increasing promise to develop more targeted, less invasive therapies for T1D treatment. Towards this goal, researchers at IBM and Johns Hopkins University (JHU) School of Medicine have discovered a unique type of cell set that induces a T1D autoimmune reaction—the first evidence of such a reaction ever reported. The discovery, published in Cell, could give us clues into the root causes of T1D.

In T1D, immune cells called “killer” T-cells receive signals from “helper” T-cells to destroy beta cells. Beta cells are cells within the pancreas that produce insulin. Previously, researchers have investigated the possibility that insulin-producing beta cells may be encouraging their own destruction by sending signals to helper T-cells to carry out this attack1. Unfortunately, the large effort invested in uncovering these mechanisms has yielded confounding results2.

In the new study, researchers at JHU and IBM looked beyond insulin-producing beta cells and revealed a novel and previously undiscovered T1D-causing immune cell set, termed dual expressor (DE) cells. These cells are so named because they express both immune proteins, T-cell receptor and B-cell receptor, which have never been observed in the same cell.

Importantly, a portion of one of these receptors, the B-cell receptor, was found to prime helper T-cells to stimulate killer T-cells to destroy insulin-producing beta cells. This is intriguing, because it hints at a new, undiscovered possible response from the immune system. Instead of a direct attack on insulin-producing beta cells in the pancreas, this suggests another autoimmune response towards the pancreas driven by a reaction from other immune cells (DE cells).

Initially, researchers at JHU identified expanded populations of DE cells in the blood of individuals with T1D but did not know if they caused a T1D immune response. Our team of researchers in the IBM Healthcare and Life Science division worked closely with the experimental collaborators to determine that a specific peptide of the DE cells (autoantigen) did indeed cause a T1D immune response.

IBM researchers used large-scale computer simulations to model the binding of the autoantigen peptides from DE cells to T1D-associated specific immune proteins, Human Leukocyte Antigen proteins (HLA-DQ8). The autoantigen from DE cells was found to bind to HLA-DQ8 molecules over ~10,000 times more strongly than insulin. Further, IBM researchers revealed an unusual binding behavior, with almost perfect binding registry, for the autoantigen to HLA-DQ8, explaining why the autoantigen produces a strong immune response despite having a different protein sequence from insulin.

(A) Killer T-cells recognize autoantigens presented by the HLA protein of helper T-cells, and destroy insulin producing islet β-cells in the pancreas. The image was adapted from [Credit: Carla Schaffer / M.A. Ali et al., / AAAS, https://medicalxpress.com/news/2017-08-immunotherapy-diabetes.html ]. (B) A representative structure of HLA presenting the IGHV04-B antigen. (C) Overlay of the structures of IGHV04-B (blue) and superagonist (red) at the binding groove of HLA. (D) Relative HLA binding affinity of IGHV04-B, superagonist, and insulin. The binding affinity is the highest for IGHV04-B, and lower in turn for the superagonist and insulin.

The results of this work may be valuable to the design of potential immunotherapies for T1D. Specifically, potential vaccinations could be designed so that DE cells are marked harmful and destroyed by the immune system. Likewise, novel therapies could be designed so that DE cells do not elicit a helper T-cell response or stimulate the immune system. Although these solutions require additional investigation, as potential effectual and causal treatments for T1D, they could offer substantial improvement in T1D treatment.

At IBM, we are exploring how the results of this research can be applied to advance the potential of T1D therapeutics, and this is another step forward in our mission to help alleviate the burden of diabetes with technology.


  1. https://www.pnas.org/content/115/1/162.long
  2. https://academic.oup.com/intimm/article/21/6/705/702452

Distinguished RSM and Manager, IBM Research

More Healthcare stories

What’s in your food? A new technology platform shows early signs of promise

Food authentication is emerging as an important field, and at IBM Research we are applying a multidisciplinary approach in collaboration with industry and academia to address this global challenge.

Continue reading

Meet Ari, the Smart Bike that Helps you Catch Green Lights

As more bikes take to the city streets than ever, researchers at RMIT have designed Ari, a smart e-bike that could help riders cruise the ‘green wave’.

Continue reading

Dark Matter Matters: AI Makes DNA Dark Matter Useful

What is the minimal description that captures a space? Asking a mathematician’s basic question of a  biological dataset reveals interesting answers about biology itself. This summarizes our underlying approach to subtyping hematological cancer. Disease subtyping is a central tenet of precision medicine, and is the challenging task of identifying and classifying patients with similar presentations […]

Continue reading