Nature CommunicationsPhysiologyPeer-reviewed journal article
Heterogeneous endocrine cell composition defines human islet functional phenotypes
Carmella Evans‐MolinaiD, Yasminye D. PettwayiD, Diane C. SaundersiD, Seth A. SharpiD, Thomas S. R. BateiD, Han SuniD, Heather H. Durai, Shaojun MeiiD, Anastasia Coldren, Corey Davis, Conrad ReihsmanniD, Alexander L. HopkirkiD, Jay TayloriD, Amber M. BradleyiD, Radhika AramandlaiD, Greg PoffenbergeriD, Adel EskarosiD, Regina Jenkins, Danni Shi, Ke XuiD, Hakmook KangiD, Varsha Rajesh, Swaraj ThamaniD, Fan FengiD, Jean‐Philippe CartailleriD, , Marcela BriššováiD
Abstract
Phenotyping and genotyping initiatives within the Integrated Islet Distribution Program (IIDP), the largest source of human islets for research in the U.S., provide standardized assessment of islet preparations distributed to researchers, enabling the integration of multiple data types. Data from islets of the first 299 organ donors without diabetes, analyzed using this pipeline, highlights substantial heterogeneity in islet cell composition associated with hormone secretory traits, sex, reported race and ethnicity, genetically predicted ancestry, and genetic risk for type 2 diabetes (T2D). While and {beta} cell composition influenced insulin and glucagon secretory traits, the abundance of {delta} cells showed the strongest association with insulin secretion and was also associated with the genetic risk score (GRS) for T2D. These findings have important implications for understanding mechanisms underlying diabetes heterogeneity and islet dysfunction and may provide insight into strategies for personalized medicine and {beta} cell replacement therapy.
- organism: human
- assay: imaging
- assay: scRNA-seq
- assay: bulk RNA-seq
- research focus: polygenic risk
- research focus: database / resource
- research focus: differential expression
- research focus: cell type annotation
- approach: statistical model
- approach: experimental assay
In-depth summary
Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 2 (25 Aug 2025) · CC BY-NC-ND 4.0
In 299 islet preparations from donors without diabetes, endocrine cell composition, not donor demographics, tracked insulin secretion, and δ cell percentage rose with type 2 diabetes genetic risk.
Mouse and human islets differ in function and architecture, and earlier studies disagree on whether donor age or BMI alters in vitro insulin secretion. This work combines standardized perifusion, histological cell counts and genotyping in one larger donor set while controlling for islet processing, and the authors describe it as, to their knowledge, the first direct comparison of measured islet composition with hormone secretion.
Key results
ShownEndocrine cell composition, rather than donor traits or islet morphology, tracked insulin secretion. A higher β cell percentage went with more stimulated insulin release; higher α and δ cell percentages went with less. Islet diameter, area and perimeter showed no association with any secretion trait.
Scaled coefficients in eight-covariate regressions (n = 299): β cell % vs glucose-stimulated b = 0.23 (p = 2.31 x 10^−4), cAMP-potentiated b = 0.35 (p = 1.18 x 10^−8) and KCl-mediated insulin secretion b = 0.30 (p = 1.16 x 10^−6); δ cell % vs glucose stimulation index b = –0.13 (p = 0.02) and low glucose plus adrenaline inhibition index b = –0.19 (p = 1.76×10^−3). · Islet composition strongly influences hormone…; DISCUSSION; Figure legends
SuggestedA type 2 diabetes (T2D) genetic risk score (GRS) built from 338 variants was positively associated with islet δ cell percentage, whether or not HbA1c was included as a covariate.
268 genotyped donors; covariates were age, sex, BMI, isolation center, culture and transit times and five genetic principal components. b = 0.17, p = 0.019 with HbA1c; b = 0.17, p = 0.016 without. Without HbA1c, the T2D GRS was also positively associated with glucagon content. With HbA1c included, no other T1D or T2D GRS association with secretion or content was significant. · Genetic risk for diabetes is associated with…; EXTENDED DATA FIGURE AND TABLE LEGENDS; Figure legends
SuggestedIn Human Pancreas Analysis Program (HPAP) islet scRNA-seq, genes nearest the T2D GRS variants were enriched among genes upregulated in δ cells. HHEX, DGKB and GLP1R were among the GRS genes expressed in δ cells.
87,505 islet cells (1877 δ cells) from 44 donors without diabetes. The 338 variants mapped to 303 nearest genes; 181 of 303 (60%) were expressed in δ cells at mean normalized counts > 0.01. Tests were Wilcoxon rank-sum in scanpy with Benjamini-Hochberg correction, plus a Fisher test for enrichment; the text prints no enrichment statistic. · Genetic risk for diabetes is associated with…; DISCUSSION; Differentiational expression of predicted T2D…; Figure legends
ShownEndocrine composition varied widely between donors and differed by sex and ancestry. Female donors had more α and fewer β cells, and Asian or East Asian donors had more β cells. Age, BMI, HbA1c, isolation center and processing times had no detectable effect on composition.
Mean 58% β (range 25 – 92%), 34% α (3 – 68%) and 8% δ (1 – 19%), roughly 4-, 22- and 19-fold variation. Global effects on β and α percentages: reported race or ethnicity (p = 0.003), genetic ancestry (p = 0.0004). Asian vs White and East Asian vs European donors had more β and fewer α cells (p = 0.0007 and p = 0.0001), based on 12 Asian and 11 East Asian donors. · Islet composition is influenced by sex, reported…; DISCUSSION; Figure legends
ShownAfter adjusting for covariates and multiple testing, donor age, sex and reported race or ethnicity showed no association with secretion. BMI was linked to glucagon traits, predicted ancestry to basal insulin, and pre-shipment culture and transit times to several insulin and glucagon traits.
BMI coefficients: basal glucagon b = –0.20 (p = 0.005), KCl-mediated glucagon b = –0.17 (p = 0.03), glucose inhibition of glucagon b = 0.15 (p = 0.03). Unadjusted Spearman links, e.g. age vs cAMP-potentiated insulin (r = –0.15), were not retained after adjustment. Secretion did not differ between donors with HbA1c < 5.7% (n = 223) and ≥ 5.7% (n = 76). · Islet secretory function is highly heterogeneous…; DISCUSSION; EXTENDED DATA FIGURE AND TABLE LEGENDS
Methods, data and reproducibility
Islets from 299 donors without diabetes were phenotyped (purity, viability, perifusion secretion as 11 insulin and 9 glucagon traits, hormone content, immunofluorescence cell composition) and genotyped (Omni2.5Exome array, TOPMed imputation, T1D/T2D genetic risk scores, ancestry). Associations used scaled multivariable regression with false discovery rate control.
- Integrated Islet Distribution Program (IIDP) islet preparations from organ donors without diabetes, phenotyped by the Human Islet Phenotyping Program (5 isolation centers, 2016-2024) · human · n = 299 (223 with HbA1c < 5.7%, 76 with HbA1c ≥ 5.7%)
- Donor genotypes (Illumina Infinium Omni2.5Exome V1.5, imputed against TOPMed R3 on GRCh38) used for genetic risk scores and ancestry prediction · human · n = 268 analysed (269 samples after QC and imputation) · EGAS50000000697
- Re-processed HPAP islet scRNA-seq from donors without diabetes, downloaded from PanKbase · human (Human Pancreas Analysis Program organ donors) · n = 87,505 cells (1877 δ cells) from 44 donors
- 1000 Genomes reference panel used to train the ancestry classifier (first 20 principal components) · human (1000 Genomes reference panel)
- Methods & tools
- cellSens Count and Measure (Olympus), Halo tissue classifier (Indica Labs), ScanScope FL whole slide scanner (Aperio/Leica), Illumina Infinium Omni2.5Exome V1.5 array, TOPMed R3 imputation reference panel, Random forest ancestry classifier (GeneticAncestry), BEDTools, Ensembl v112 annotation, scanpy v1.10.4, R 4.3.0, GraphPad Prism 10, Insulin and glucagon ELISA, RIA and HTRF kits (Mercodia, Millipore, R&D Systems, Cisbio)
- Data & accessions
- https://iidp.coh.org/Resources-Offered/Research-Data-Repositoryhttps://pancreatlas.org/datasets/853/explorehttps://zenodo.org/records/15588240https://pancreatlas.org/datasets/1138/explore/omics/17https://powersbrissovalab.shinyapps.io/scRNAseq-Islets/
- Software & environment
- Standardized protocols at https://iidp.coh.org/SOPs; R 4.3.0, GraphPad Prism 10, scanpy v1.10.4, Ensembl v112; imputation against TOPMed R3 on GRCh38. No container or requirements file is stated.
- Reproducibility gaps
- Genotypes in EGA (EGAS50000000697) require a data access agreement. Per-donor data are in Supplemental File 2 and the IIDP Research Data Repository. The authors provide any further information needed for reanalysis on request.
Limitations (6)
- Islets came from deceased organ donors after brain or circulatory death, and the reported cause of death, which may affect secretory function, was not analysed. · stated by the authors
- Secretion was tested only with glucose-based stimuli plus IBMX, adrenaline and KCl. Responses to fatty acids or amino acids were not examined. · stated by the authors
- Several associations were only nominally significant, and the authors say larger samples are needed. · stated by the authors
- Few donors were reported as Asian (n = 12), so their higher β cell percentage needs confirmation in larger samples. · stated by the authors
- The link between the T2D GRS and δ cell percentage comes from one cohort of 268 donors and was not replicated externally. The scRNA-seq support assigns variants to their nearest genes rather than to known effector transcripts. · noted in this summary
- Hormone assays switched between RIA, HTRF and ELISA kits over 2016-2024, and neither assay platform nor islet purity was a covariate in the regression models. · noted in this summary
Read this if you run experiments on donor human islets and need to know which donor, processing and composition variables shape secretion, or if you study δ cells in type 2 diabetes genetics.
- Citations
- 12
- FWCI
- 60.18
- Field percentile
- top 1%
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- HumanIslets: An integrated platform for human islet data access and analysis
Jessica Ewald, Yao Lu, … Patrick E. MacDonald·bioRxiv·Jun 19, 2024
- Islet size, cellular composition, and spatial location are interconnected in the human pancreas
Sean D. Dufault-Hunter, Vipul Periwal, Manami Hara·bioRxiv·Oct 8, 2026
- Single-cell decoding of human islet cell type-specific alterations in type 2 diabetes reveals converging genetic- and state-driven β-cell gene expression defects
Khushdeep Bandesh, Efthymios Motakis, … Michael L. Stitzel·The EMBO JournalPeer-reviewed·Apr 15, 2026
- AI-guided analysis of human pancreatic islet sociology reveals distinct cell compositional changes in type 1 diabetes
Caroline Ward, Tabitha Banks‐Tibbs, … Zachary Freyberg·bioRxiv·Jun 9, 2026
- Single cell regulatory architecture of human pancreatic islets suggests sex differences in β cell function and the pathogenesis of type 2 diabetes.
Mirza Muhammad Fahd Qadir, Ruth M. Elgamal, … Franck Mauvais‐Jarvis·The EMBO JournalPeer-reviewed·Nov 20, 2024
- Sex differences in islet stress responses support female beta cell resilience
G. P. Brownrigg, Y. H. Xia, … E. J. Rideout·bioRxiv·May 10, 2022