Early Mechanisms of Chemoresistance in Retinoblastoma.

Retinoblastoma is the most common eye cancer in children and is fatal if left untreated. Over the past three decades, chemotherapy has become the mainstay of eye-sparing treatment. Nevertheless, chemoresistance continues to represent a major challenge leading to ocular and systemic toxicity, vision loss, and treatment failure. Unfortunately, the mechanisms leading to chemoresistance remain incompletely understood. Here, we engineered low-passage human retinoblastoma cells to study the early molecular mechanisms leading to resistance to carboplatin, one of the most widely used agents for treating retinoblastoma. Using single-cell next-generation RNA sequencing (scRNA-seq) and single-cell barcoding technologies, we found that carboplatin induced rapid transcriptomic reprogramming associated with the upregulation of PI3K-AKT pathway targets, including ABC transporters and metabolic regulators. Several of these targets are amenable to pharmacologic inhibition, which may reduce the emergence of chemoresistance. We provide evidence to support this hypothesis using a third-generation inhibitor of the ABCB1 transporter.

BAP1 Loss Promotes Suppressive Tumor Immune Microenvironment via Upregulation of PROS1 in Class 2 Uveal Melanomas.

Uveal melanoma (UM) is the most common primary cancer of the eye and is associated with a high rate of metastatic death. UM can be stratified into two main classes based on metastatic risk, with class 1 UM having a low metastatic risk and class 2 UM having a high metastatic risk. Class 2 UM have a distinctive genomic, transcriptomic, histopathologic, and clinical phenotype characterized by biallelic inactivation of the BAP1 tumor-suppressor gene, an immune-suppressive microenvironment enriched for M2-polarized macrophages, and poor response to checkpoint-inhibitor immunotherapy. To identify potential mechanistic links between BAP1 loss and immune suppression in class 2 UM, we performed an integrated analysis of UM samples, as well as genetically engineered UM cell lines and uveal melanocytes (UMC). Using RNA sequencing (RNA-seq), we found that the most highly upregulated gene associated with BAP1 loss across these datasets was PROS1, which encodes a ligand that triggers phosphorylation and activation of the immunosuppressive macrophage receptor MERTK. The inverse association between BAP1 and PROS1 in class 2 UM was confirmed by single-cell RNA-seq, which also revealed that MERTK was upregulated in CD163+ macrophages in class 2 UM. Using ChIP-seq, BAP1 knockdown in UM cells resulted in an accumulation of H3K27ac at the PROS1 locus, suggesting epigenetic regulation of PROS1 by BAP1. Phosphorylation of MERTK in RAW 264.7 monocyte-macrophage cells was increased upon coculture with BAP1-/- UMCs, and this phosphorylation was blocked by depletion of PROS1 in the UMCs. These findings were corroborated by multicolor immunohistochemistry, where class 2/BAP1-mutant UMs demonstrated increased PROS1 expression in tumor cells and increased MERTK phosphorylation in CD163+ macrophages compared with class 1/BAP1-wildtype UMs. Taken together, these findings provide a mechanistic link between BAP1 loss and the suppression of the tumor immune microenvironment in class 2 UMs, and they implicate the PROS1-MERTK pathway as a potential target for immunotherapy in UM.

Uphyloplot2: visualizing phylogenetic trees from single-cell RNA-seq data.

BACKGROUND: Recent advances in single cell sequencing technologies allow for greater resolution in assessing tumor clonality using chromosome copy number variations (CNVs). While single cell DNA sequencing technologies are ideal to identify tumor sub-clones, they remain expensive and in contrast to single cell RNA-seq (scRNA-seq) methods are more limited in the data they generate. However, CNV data can be inferred from scRNA-seq and bulk RNA-seq, for which several tools have been developed, including inferCNV, CaSpER, and HoneyBADGER. Inferences regarding tumor clonality from CNV data (and other sources) are frequently visualized using phylogenetic plots, which previously required time-consuming and error-prone, manual analysis. RESULTS: Here, we present Uphyloplot2, a python script that generates phylogenetic plots directly from inferred RNA-seq data, or any Newick formatted dendrogram file. The tool is publicly available at https://github.com/harbourlab/UPhyloplot2/ . CONCLUSIONS: Uphyloplot2 is an easy-to-use tool to generate phylogenetic plots to depict tumor clonality from scRNA-seq data and other sources.

PieParty: visualizing cells from scRNA-seq data as pie charts.

Single-cell RNA sequencing (scRNA-seq) has been a transformative technology in many research fields. Dimensional reduction techniques such as UMAP and tSNE are used to visualize scRNA-seq data in two or three dimensions for cells to be clustered in biologically meaningful ways. Subsequently, gene expression is frequently mapped onto these plots to show the distribution of gene expression across the plots, for instance to distinguish cell types. However, plotting each cell with only a single color leads to repetitive and unintuitive representations. Here, we present PieParty, which allows scRNA-seq data to be plotted such that every cell is represented as a pie chart, and every slice in the pie charts corresponds to the gene expression of a single gene. This allows for the simultaneous visualization of the expression of multiple genes and gene networks. The resulting figures are information dense, space efficient, and highly intuitive. PieParty is publicly available on GitHub at https://github.com/harbourlab/PieParty.