Endocrine Abstracts

Context: Pathological transcriptomic landscape has been the driver of the Molecular Biology in the recent years. Massive RNA sequencing techniques have been extended from bulk RNAseq, to single cell sequencing (scRNAseq) and spot-based spatial transcriptomics (ST), and sometimes laboratory cannot afford all of them. We tried to answer the question about which would be the best option in thyroid gland using Hashimoto’s thyroiditis (HT) context.

Methodology: We performed RNAseq (5 HT, 5 controls) and Visium ST using the 10XGenomics platform (3 HT, 2 controls). We used scRNAseq from four HT patients from a public repository database. For ST, we followed a histological annotation classification and we analyzed each tissue compartment: thyrocytes, connective tissue, vessels, germinal center (GC) and immune infiltrated cells (TILs), separately. All the analysis were executed in R and Seurat package. DESeq was used for differential expression analysis (DEA) in RNAseq. In scRNAseq and ST, DEA was performed under the default option of Seurat.

Results: We first focused our analysis on the ST data separating the different tissue compartments. We found two thyrocytes subpopulations (“healthy” and “damaged”), three main fibroblasts subpopulations and a specific marker of vessels permeabilization, PLVAP, with their respective locations. Furthermore, we observed the spatial concentration of dendritic cell markers (such as FDCSP) in GCs, a main characteristic of HT histology. Nevertheless, TILs surrounding the GC were not properly estimated nor classified. In scRNAseq, we characterized different immune subpopulations which were not able to be observed in ST. However, it is critical to consider the low number of thyrocytes compared to other cells which smggests a non-uniform coverage of all representative cells in the thyroid tissue. This fact could interfere with the results of the analysis presenting limitations in the description of “damaged” and “healthy” thyrocytes subpopulations. We also found a lack of expression of FDCSP marker in GC cells represents the low capture efficiency or the damaged cells that results in the scRNAseq. In RNAseq, the upregulation of some genes is overshadowed by immune cell markers. In the top, we found genes such as CXCL13 and FDCSP. The analysis of pathogenic-fibroblasts markers and PLVAP did not reach statistical significance. It may be caused by the lack of depth/samples in the RNAseq analysis and the overrepresentation of immune cells transcriptomes.

Conclusions: ST provides a robust method to identify different cell type subpopulations similar that in scRNAseq in non-immune cells and offers higher depth than RNAseq.