Spatial RNA Sequencing (spRNA-seq, 10x Genomics, Visium)

FFPE samples were selected based on their size, time of storage, and if available the previous report of increased Ki-67 index as a marked of increased cell proliferation. FFPE samples were cut and sectioned directly onto the 10x Genomics barcoded Visium slides (10X genomics part# 1000338). Two Visium slides were used, each containing four capture areas that were 6.5 × 6.5mm². Every capture area consisted of approximately 5,000 gene expression spots with primers for sequencing and identification in downstream analyses. Seven of the eight panels contained tumor samples used for this study. Six of the seven total capture areas contained a single pituitary tumor. The remaining capture area had two tissues: a pituitary tumor and a small piece of normal gland used as a control.

All steps from 10X Genomics’ FFPE Visium Spatial workflow-specific protocols were followed. In short, after the slices were placed on the slides, they were then deparaffinized, stained with Hematoxylin & Eosin (H&E), imaged, and decrosslinked according to the Demonstrated Protocol (10x Genomics- protocol {"type":"entrez-nucleotide","attrs":{"text":"CG000409","term_id":"33869057","term_text":"CG000409"}}CG000409 Rev B). Samples were processed using the Visium Spatial Gene Expression Reagent Kits for FFPE (10x Genomics – protocol {"type":"entrez-nucleotide","attrs":{"text":"CG000407","term_id":"33869055","term_text":"CG000407"}}CG000407 Rev D). A human whole transcriptome probe panel was used to hybridize probes to their complementary RNA on the tissue sections. The probe pairs were ligated together and released from the tissue through RNAse treatment and permeabilization before getting captured onto the Visium Slide. The ligation products were extended and consisted of partial read 1, 10X Genomics spatial barcode, unique molecular identifier, Poly A, probe insert, and partial read 2 sequences. The ligated probe products underwent a sample index PCR to add P5, i5, i7, and P7 sequences needed for the Illumina platform. Sequencing was performed using the Illumina NovaSeq 6000 at a minimum of 25,000 read pairs per tissue-covered spot on the capture area. The sequencing saturation was above 40% for all samples with a median of 91,447 reads and 6,494 expressed genes per spot of tissue.

All differential spatial sequencing analyses were conducted using the Cell Ranger analysis pipeline. Principal component analysis (PCA) was run on the feature barcode matrices and clustering was reported based on the minimum number of gene dimensions. Results, including specific gene or gene list expression profiles and uniform manifold approximation and projection (UMAP) were reviewed at the 10X Genomics Loupe Browser. Lists of select marker genes used in certain analyses (marker genes of corticotroph cells, non-corticotroph pituitary lineages, collagen and fibroblasts, and genes of p53 and MAPK pathway as listed in Wiki pathway database) are provided in supplementary material. Gene pathway analysis was performed on differentially expressed genes for each slide in R using the clusterprofiler package for gene set enrichment analysis. Wiki, Hallmark and KEGG pathways were reviewed to identify enriched pathways within each cluster and identify possible pathways of interests involved in these processes.