5-Fluorouracil treatment represses pseudouridine-containing miRNA export into extracellular vesicles

Abstract

5-Fluorouracil (5-FU) has been used for chemotherapy for colorectal and other cancers for over 50 years. The prevailing view of its mechanism of action is inhibition of thymidine synthase leading to defects in DNA replication and repair. However, 5-FU is also incorporated into RNA causing defects in RNA metabolism, inhibition of pseudouridine modification, and altered ribosome function. We examined the impact of 5-FU on post-transcriptional small RNA modifications (PTxMs) and the expression and export of RNA into small extracellular vesicles (sEVs). EVs are secreted by all cells and contain a variety of proteins and RNAs that can function in cell-cell communication. We found that treatment of colorectal cancer (CRC) cells with 5-FU represses sEV export of miRNA and snRNA-derived RNAs, but promotes export of snoRNA-derived RNAs. Strikingly, 5-FU treatment significantly decreased the levels of pseudouridine on both cellular and sEV small RNA profiles. In contrast, 5-FU exposure led to increased levels of cellular small RNAs containing a variety of methyl-modified bases. These unexpected findings show that 5-FU exposure leads to altered RNA expression, base modification, and aberrant trafficking and localization of small RNAs.

Keywords: 5‐FU; EV export; RNA modification; extracellular vesicles; miRNA; pseudourine.

FIGURE 1

Differential gene expression in DLD‐1 cells after exposure to 5‐FU. DLD‐1 cells were exposed or not to 10 uM 5‐FU for 48 h in the absence of serum after which RNA was isolated and RNAseq was performed on transcripts > 200 nt. (a) Volcano plot showing up‐ and down‐regulated genes after treatment with 5‐FU. Blue dots show downregulated genes and red dots show upregulated genes with at least 2‐fold changes in expression and adjusted p values < 0.05. Grey dots represent transcripts whose expression did not significantly change after 5‐FU treatment. (b) Hierarchical clustering analysis was performed on triplicate RNAseq samples prepared from DLD‐1 cells treated (red) or not (blue) with 5‐FU. The heat map shows up‐ (red) and down‐ (blue) regulated genes. (c) Gene ontology analysis was performed using the WebGestaltR package to identify overlaps between gene annotation sets after exposure to 5‐FU (Wang et al., 2017).

FIGURE 2

Small RNAseq of DLD‐1 cells and sEVs treated with 5‐FU. Small RNAseq was performed on RNA (<200 nt) isolated from DLD‐1 cells and sEVs treated or not with 5‐FU. (a) Small RNA read totals normalized to reads per million (RPM) from the indicated RNA subclasses after RNAseq on cellular and sEV RNA isolated in the presence or absence of 5‐FU. RNA subclasses include long noncoding RNA‐derived RNAs (lncDRs), miRNAs, miscellaneous small RNAs, rRNA‐derived RNAs (rDRs), spliceosomal snRNA‐derived RNAs (snDRs), snoRNA‐derived RNAs (snoDRs), tRNA‐derived RNAs (tDRs) and Y RNA‐derived RNAs (yDRs). (b) Distinct clustering of miRNA expression patterns across triplicate RNAseq samples in cells and sEVs after treatment with 5‐FU.

FIGURE 3

Differentially expressed small RNAs comparing DLD‐1 cells and sEVs grown in the presence or absence of 5‐FU. Volcano plots showing up‐ and down‐regulated miRNAs (a), snDRs (b) and snoDRs (c) in cells and sEVs treated or not with 5‐FU. Blue (downregulated) and red (upregulated) dots represent individual RNAs with significant changes in gene expression after treatment with 5‐FU (padj < 0.05 and fold‐change >1.5 or < 0.67).

FIGURE 4

Increased uridine content in cellular‐retained miRNAs after 5‐FU treatment. (a) miRNAs that display coordinate up‐regulation in DLD‐1 cells and down‐regulation in sEVs after treatment with 5‐FU are shown with the number of uridine residues in each mature miRNA. miR‐340 and miR‐92a also contain uridine rich precursor loop sequences. (b) Overall enrichment of uridine residues in miRNAs retained in cells after 5‐FU treatment. The percentage of U residues in cellular retained miRNAs was compared between those that were significantly enriched (sig) or not (nsig) in cells after 5‐FU exposure using the Wilcoxon rank sum test. A significant increase (p < 0.001) in uridine content was detected in cellular‐retained RNAs with the average and standard deviation as indicated.

FIGURE 5

Reduced pseudouridine modification in cells and sEVs after 5‐FU treatment. LC–MS/MS approaches were used to quantify RNA base modifications in RNA from cells and sEVs. (a) Nucleoside content in 5‐FU treated DLD‐1 cells showed a modest but significant increase (p < 0.05) in uridine content when compared to untreated cells. No changes were observed for the other RNA nucleosides. (b) No changes in nucleoside levels were detected in sEVs purified from cells treated or not with 5‐FU. (c) Pseudouridine levels were significantly decreased in both cells and sEVs after 5‐FU treatment. (d−l) Quantitation of the effect of the indicated base modifications in cells and sEVs treated or not with 5‐FU. Base modifications include dihydrouridine (DHU), methylated cytosine (m⁵C and m³C), methylated adenosine (m¹A), methylated inosine (m¹I), and methylated guanosine (m¹G, m⁷G, m²G) and demethylated guanosine (m² ₂G). To compare modification levels across conditions, modified nucleoside values were normalized to the total levels for each nucleoside. For all graphs, the error bars indicate average and standard deviation.