Abstract: SA-PO0067
Spatial Immune-Tubular Crosstalk in Diabetic Kidney Disease: Proximity to Immune Cells Drives Proximal Tubule Injury Gene Programs
Session Information
- Artificial Intelligence: Kidney Imaging, Pathology, and Precision Phenotyping
October 24, 2026 | Location: Exhibit Hall A, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- D'Antonio, Matteo, University of Washington, Seattle, Washington, United States
- Lee, Da Young, Korea University College of Medicine, Seongbuk-gu, Seoul, Korea (the Republic of)
- Choi, Ye Ji, University of Washington, Seattle, Washington, United States
- Weissenkampen, James Dylan, University of Washington, Seattle, Washington, United States
- Miller, Kristen, University of Washington, Seattle, Washington, United States
- Ramesh, Shivani, University of Washington, Seattle, Washington, United States
- Leidholt, Savanah L., University of Washington, Seattle, Washington, United States
- Pinzon Cortes, Jairo Arturo, University of Washington, Seattle, Washington, United States
- Kendrick, Jessica B., University of Colorado Anschutz Medical Campus, Aurora, Colorado, United States
- Teitelbaum, Isaac, UCHealth University of Colorado Hospital, Aurora, Colorado, United States
- Smith, Cathy, University of Michigan Michigan Medicine, Ann Arbor, Michigan, United States
- Driscoll, Lynette M., University of Colorado Anschutz Medical Campus, Aurora, Colorado, United States
- Hodgin, Jeffrey B., University of Michigan Michigan Medicine, Ann Arbor, Michigan, United States
- McCown, Phillip J., University of Michigan Michigan Medicine, Ann Arbor, Michigan, United States
- Nair, Viji, University of Michigan Michigan Medicine, Ann Arbor, Michigan, United States
- Vaisar, Tomas, University of Washington, Seattle, Washington, United States
- Kanter, Jenny E., University of Washington, Seattle, Washington, United States
- Bornfeldt, Karin E., University of Washington, Seattle, Washington, United States
- Goodrich, Jesse Allen, University of Southern California Keck School of Medicine, Los Angeles, California, United States
- Karihaloo, Anil K., Research Science Center, Novo Nordisk, Lexington, Massachusetts, United States
- Tommerdahl, Kalie L., University of Washington, Seattle, Washington, United States
- Pyle, Laura, University of Washington, Seattle, Washington, United States
- Bjornstad, Petter, University of Washington, Seattle, Washington, United States
- Kretzler, Matthias, University of Michigan Michigan Medicine, Ann Arbor, Michigan, United States
- Kim, Nan Hee, Korea University Ansan Hospital, Ansan-si, Gyeonggi-do, Korea (the Republic of)
Background
Tubulointerstitial inflammation is a key driver of diabetic kidney disease (DKD) progression, yet the spatial effects of immune infiltration on proximal tubules (PT) have been poorly characterized at single cell resolution. We hypothesized that PT cells adjacent to immune infiltrates express injury-associated genes, while those distant from immune cells retain markers of healthy, differentiated epithelium.
Methods
We analyzed Xenium 5000 spatial transcriptomic data from kidney biopsies of 31 participants (3 healthy controls, 28 with type 2 diabetes). Cell types were annotated through label transfer from the Kidney Precision Medicine Project. Immune spatial architecture was assessed via Tertiary Lymphoid Structure (TLS) detection. TLS were identified by spatially clustering lymphoid cells with DBSCAN (75 µm radius, minimum 15 cells). We computed the distance from each PT cell to its closest immune cell and tested the association between this distance and PT gene expression using linear mixed models.
Results
While healthy controls did not display any TLS, T2D samples had 4.9±4.3 TLS (p = 0.041, Kruskal-Wallis). Participants with T2D also exhibited increased immune cells, fibroblast. Gene expression modeling identified HAVCR1 as the top gene positively associated with immune proximity in PT cells, while SLC22A8 (OAT3), a canonical transporter of differentiated PT epithelium, was among the top negatively associated genes.
Conclusion
Spatial transcriptomics reveals that immune-PT proximity in DKD is linked to a local injury transcriptional program. PT cells closest to immune infiltrates exhibit an injury/dedifferentiation phenotype, with HAVCR1 upregulation and downregulation of differentiation markers such as SLC22A8. In contrast, PT cells that are distant from immune cells retain functional transporter expression, providing spatial evidence that immune-tubular crosstalk drives localized tubular dedifferentiation in DKD.
Expression of HAVCR1 and SLC22A8 in PTs: HAVCR1 (red, left) is expressed in PTs in close proximity to immune cells, while SLC22A8 (orange, right) is expressed in PTs not affected by immune infiltration,
Funding
- Commercial Support – Eli Lilly, AstraZeneca, Jansen, Novo Nordisk and Gilead