Potential Improvements to Diabetes Drug Discovery and Development

Research that tackles the issues posed by current islet screening methods, produces a high throughput bioassay with a higher relevance than other methods.

Background

Diabetes poses a global health crisis that affects individuals across age groups and backgrounds. Diabetes and consequential conditions affect 462 million people worldwide. Approximately 12% of global health expenditure is dedicated to diabetes and associated conditions. The prevalence is predicted to increase as countries become more developed and it is estimated that one in ten adults, globally, will be affected by diabetes by 2040. Diabetes mellitus represents a spectrum of persistent pathological conditions initiated by disruption in blood glucose homeostasis—the self-regulating process of maintaining internal stability despite external dynamics. Current therapeutic strategies primarily rely on insulin therapy or hypoglycemic agents, which fail to address the root cause of the disease—the loss of pancreatic insulin-producing beta-cells. Therefore, bioassays that recapitulate intact islets are needed to enable drug discovery for beta-cell replenishment, protection from beta-cell loss, and islet-cell interactions. Standard cancer bioassays are not suitable for studying proliferative effects. Screening using primary human or rodent intact islets offers a higher level of physiological relevance to enhance diabetes drug discovery and development. However, the three-dimensional nature of intact islets has presented challenges in developing robust, high-throughput assays to detect beta-cell proliferative effects. Established methods rely on either dissociated islet cells plated in 2D monolayer cultures for imaging or reconstituted pseudo-islets formed in round bottom plates to achieve homogeneity. These approaches have significant limitations due to the islet cell dispersion process

High-Content Imaging Assay

To address the limitations described, above, a research team from the College of Pharmacy, University of Michigan Medical School, and University of Michigan Center for Drug Repurposing, all in Ann Arbor, Michigan, has developed a robust, intact ex vivo pancreatic islet bioassay that is capable of detecting diabetes-relevant endpoints including beta-cell proliferation, chemoprotection, and islet spatial morphometrics.

Figure 1. Workflow for high-content imaging of intact islets. (1) Islets are harvested from mice and placed in 384-well plates. (2) After fixation, islets are stained with Hoechst, Ki67, Nkx6.1, and Glucagon antibodies. (3) Imaging on a Yokogawa CV8000 with a 4X overview image of the entire well followed by centered 40 × 3D-imaging of each islet. (4) Images are analyzed with Cellpose and CellProfiler for islet and cell segmentation and feature extraction. Data analysis is performed using Knime.

Image Acquisition

Figure 2. Image acquisition pipeline for the CV8000. A single 4x field captures the entire well for islet identification. 4X well images are then processed using CellProfiler to identify islets and filter out unwanted objects. The location of each identified islet is fed back to the microscope and imaged at 40x in second-pass 3D imaging.

Figure 3. Example montage of an islet image stack. Rows correspond to MIP slices and columns are channels. Z01 represents the bottom of the islet and Z05 is the top. Hoechst staining is in gray, Ki67 antibody staining in green, Nkx6.1 in the middle cyan column, and glucagon is represented in red in the fourth column. Column 5 contains a merge of the Ki67, Nkx6.1, and glucagon channels.

Object Identification

Figure 2. Image acquisition pipeline for the CV8000. A single 4x field captures the entire well for islet identification. 4X well images are then processed using CellProfiler to identify islets and filter out unwanted objects. The location of each identified islet is fed back to the microscope and imaged at 40x in second-pass 3D imaging.

Object Processing

Figure 4. Top: Workflow for object identification in the 40X images. Cellpose segmentation software is used to identify all of the nuclei in the Hoechst channel. In CellProfiler, fluorescence intensity is then scaled up and an islet object is identified. Bottom: Workflow for processing identified objects in CellProfiler. Masks from CellPose are imported into CellProfiler. Nuclei that are outside of the islet rejected so measurements are only taken for islet cells. The remaining nuclei are expanded by 15 pixels to include cytoplasm surrounding the nuclei to create a ‘cell’ object. Finally, the nuclear objects are subtracted from the ‘cell’ objects to create a ‘cytoplasmic’ ring object.

Conclusion

This approach circumvents many drawbacks of dispersed islet assays by preserving the islet structure. These intact islets attach firmly to standard cell culture plates. Maintaining the islet integrity permits the analysis of cell distribution in three dimensions while eliminating potential inaccuracies due to false positives from exocrine cells or cell loss during staining. The approach enables imaging of multiple islets per well, thus significantly enhancing throughput compared to single-islet-per-well or histological methods. By initially using a low magnification scan to locate each islet in the well, researchers can subsequently perform high-resolution imaging without needing to capture the empty spaces between islets. This assay offers a high throughput bioassay with higher physiological relevance than other methods, potentially improving diabetes drug discovery and development. Although the study focused on pancreatic islet phenotyping, it is broadly applicable to other 3D organoid or spheroid cell culture systems.