LSD1 promotes secretory cell specification to driveBRAFmutant colorectal cancer

Abstract

Despite the connection to distinct mucus-containing colorectal cancer (CRC) histological subtypes, the role of secretory cells, including goblet and enteroendocrine (EEC) cells, in CRC progression has been underexplored. Analysis of TCGA and single cell RNA sequencing data demonstrates that multiple secretory progenitor populations are enriched in BRAF-mutant CRC patient tumors and cell lines. Enrichment of EEC progenitors in BRAF-mutant CRC is maintained by DNA methylation and silencing of NEUROD1, a key gene required for differentiation of EECs. Mechanistically, secretory cells and the factors they secrete, such as Trefoil factor 3, are shown to promote colony formation and activation of cell survival pathways in the entire cell population. We further identify LSD1 as a critical regulator of secretory cell specification in vitro and in a colon orthotopic xenograft model, where LSD1 loss reduces tumor growth and metastasis. This work establishes EEC progenitors, in addition to goblet cells, as targetable populations in BRAF-mutant CRC and identifies LSD1 as a therapeutic target in secretory lineage-containing CRC.

Consensus molecular subtyping of colorectal cancers is influenced by goblet cell content

“…The majority of CRCs are moderately differentiated ( 6 ) and in accordance with the TCGA COAD dataset exhibit low expression of differentiated cell type markers ( 7 ). A proportion of the TCGA COAD CRCs retain high expression of differentiation markers, particularly for secretory, goblet and enteroendocrine (EEC) cells, indicating differentiation is maintained in some but not all tumors ( 7 ). This previous work further demonstrates that the mRNA expression of strong celltype marker genes can be effectively used as a surrogate for tumor cell content.…”

“…While markers for EEC progenitors were not generally enriched in CMS3 subtypes, KLK12 and HEPACAM2 were strongly elevated in the CMS3 subtype. Given the shared early lineage between EEC and goblet cells, some genes are commonly expressed in these cell types, and we have shown the EEC progenitor marker KLK12 to be additionally expressed in goblet cells ( 7 ). Interestingly, mucinous adenocarcinomas showed no association with the CMS3 subtype or with the expression of goblet cell marker genes aside from MUC2 .…”

“…CMS subtyping classification relies on a series of calculations to determine the probability that any sample likely falls into any individual classification, with the assigned classification of a given tumor or cell line being the classification with the highest probability. We have demonstrated that the lysine-specific demethylase LSD1 is required for the persistence of goblet cells in the CMS3 HT29 cell line ( 7 ). To determine if goblet cell content is an important contributor to the probability of CMS3 subtype classification, we ran our previously published RNA-sequencing datasets comparing control empty vector (EV) versus LSD1 knockdown (KD) HT29 and SW480 cells through the CMS classifier.…”

“…TCGA COAD datasets were accessed from Xena Browser, and counts were normalized as previously described ( 7 ). To establish marker genes for each cell type, human intestine marker genes were downloaded from https://panglaodb.se (stem, goblet, enterochromaffin, and enterocyte) or compiled from previously published studies, accessed on the pubmed database (transit-amplifying, and EEC progenitor) ( 9 ).…”

“…It therefore follows that if goblet cell content is highly variable in CRC, we might be able to identify a subset of CRC with high goblet cell content under one of the CMS subtype classifications. In this study we examine the expression of marker genes for goblet cells, enterocytes, intestinal stem cells, enterochromaffin cells, which are the most abundant EEC cell of the large intestine ( 8 ), and common EEC progenitors, which were recently shown to be enriched in some CRCs ( 7 ), across the different CMS subtypes. Overall, the present study establishes proof of principle for using computational methods to rapidly query goblet cell content across sequenced tumors and establishes goblet cell content as an important predictor in the assignment of the CMS3 subtype.…”