Spatial Transcriptomics Identifies Expression Signatures Specific to Lacrimal Gland Adenoid Cystic Carcinoma Cells

Abstract

Although primary tumors of the lacrimal gland are rare, adenoid cystic carcinoma (ACC) is the most common and lethal epithelial lacrimal gland malignancy. Traditional management of lacrimal gland adenoid cystic carcinoma (LGACC) involves the removal of the eye and surrounding socket contents, followed by chemoradiation. Even with this radical treatment, the 10-year survival rate for LGACC is 20% given the propensity for recurrence and metastasis. Due to the rarity of LGACC, its pathobiology is not well-understood, leading to difficulties in diagnosis, treatment, and effective management. Here, we integrate bulk RNA sequencing (RNA-seq) and spatial transcriptomics to identify a specific LGACC gene signature that can inform novel targeted therapies. Of the 3499 differentially expressed genes identified by bulk RNA-seq, the results of our spatial transcriptomic analysis reveal 15 upregulated and 12 downregulated genes that specifically arise from LGACC cells, whereas fibroblasts, reactive fibrotic tissue, and nervous and skeletal muscle account for the remaining bulk RNA-seq signature. In light of the analysis, we identified a transitional state cell or stem cell cluster. The results of the pathway analysis identified the upregulation of PI3K-Akt signaling, IL-17 signaling, and multiple other cancer pathways. This study provides insights into the molecular and cellular landscape of LGACC, which can inform new, targeted therapies to improve patient outcomes.

Keywords: lacrimal gland adenoid cystic carcinoma; rare cancer; spatial transcriptomics; transcriptomic signature.

Figure 1

RNA sequencing analysis of LGACC vs. normal lacrimal gland. (A) Heat map of top 1000 differentially expressed genes of normal control (C) samples (n = 2) vs. LGACC (T) samples (n = 5). (B) Volcano plot of logFC by −log10(FDR) showing differentially expressed genes with logFC cutoff > 2 and <−1 and FDR < 0.05. Red and blue dots represent downregulated and upregulated genes, respectively. (C) Biological process gene ontology analysis of up- and downregulated genes. X-axis represents −log10 (p-value) of gene ontology value. Downregulated gene ontologies are represented with a negative value. (D) Box plots of RNA-seq counts per million (CPM) values for the top 10 differentially expressed collagens comparing normal lacrimal gland with LGACC. Statistical significance indicated by astricks represented as follows: * p value less than 0.03, ** p value < 0.01, *** p value less than 0.001, and **** p value < 0.0001.

Figure 2

Spatial transcriptomics elucidates LGACC and surrounding tissue signatures. (A) Hematoxylin–eosin (H&E) staining of LGACC section. (B) Spatial transcriptomic clustering results overlaid on H&E stained section. (C) Heat map of top markers for each cluster based on PCA. (D) Gene ontology biological process analysis of upregulated genes for each cluster. (E) KEGG pathway analysis of upregulated genes for each cluster.

Figure 3

Determining the identity of each cell cluster in spatial transcriptomic sample. (A) Violin plots of genes specific to cell type and top expressing genes in clusters. (B) Spatial plots showing the location of each cluster named by cell type determined from analysis.

Figure 4

Characterization of cluster 3 as potential transition or cancer stem cell cluster. (A) UMAP of cell cluster identities. (B) Violin plots of gene signature of cluster 3 alone and when compared with muscle cell clusters. (C) Monocle trajectory map indicating a zoomed-in UMAP of cluster 5 (LGACC) (indicated as 1 on the graph) as the primary and cluster 3 (indicated as 2 on the graph) cells coming from the malignant cluster. (D) UMAPs of genes of interest for cluster 3 found in trajectory analysis.

Figure 5

LGACC-specific signature identified through spatial transcriptomics and bulk RNA sequencing. (A) Venn diagram representing differentially expressed genes from bulk RNA sequencing and spatial transcriptomics and the overlap found between the two. (B) Stacked violin plot of 15 overexpressed genes that are found in both bulk and spatial transcriptomics. (C–G) Immunofluorescent staining of LGACC-specific signature and surrounding tissue signatures.