Endocrine Abstracts

JOINT567

Despite the widespread adoption of single-cell profiling for transcriptome and chromatin accessibility in the pituitary gland, there is still little agreement on cell type markers and cell population annotation. This lack of consensus stems from the disparate processing and downstream analysis methods used and arbitrary parameter and naming choices. To this end, we set out to generate a uniformly pre-processed and analysed consensus pituitary atlas (CPA), using all published single-cell profiled datasets to date (180 samples). Curation uncovered 8 out of 34 peer-reviewed publications to have either erroneously described barcoding kits or mouse samples with incorrectly assigned sexes. With these issues fixed, the atlas enables querying 750,000 cells, 10-times more than previous efforts, increasing the power of these analyses. We present the identification of novel consensus cell type markers for accurate cell typing, and markers associated with cell fate decisions. Mapping of isoform gene variants, reveals differential isoform preference across cell types. Leveraging the metadata associated with samples enabled dissection of sexual dimorphism at the cellular level. Notably, somatotrophs, lactotrophs, gonadotrophs and stem cells are dimorphic both at the gene expression and chromatin levels. Transcription factor motif enrichment suggested that these changes were driven by estrogen and androgen receptor activity in females and males respectively. We then extracted the most consistently predicted ligand-receptor interactions across all non-genetically modified (wild-type) transcriptomic samples (n = 84) using 5 different algorithms and present these for the different cell types. In stem cells, top interactions included SLIT2 signalling to Lactotrophs and Thyrotrophs, autocrine FGF signalling, as well as high expression of Erbb4 receptor (targeted by NRG3 from melanotrophs). To increase reproducibility and data availability, we developed an online platform called epitome, where researchers can explore and download analysis results and processed samples associated with the CPA. This platform will allow the CPA to be continuously updated and maintained as more datasets become available, making it a useful resource for the community.