Fig 1.
Argonaute crosslinking, ligation, and sequencing of hybrids in colorectal cancer cells.
(A) HCT116 cell lines containing knockouts for core miRNA processing machinery allow the enrichment of Drosha-independent miRNAs (e.g., mirtrons, TSS-miRNAs). (B) Schematic of AGO-CLASH. In both CLASH and qCLASH, endogenous miRNA and mRNA are crosslinked (yellow crosses) and immunoprecipitated using AGO antibody. Following AGO-IP, the miRNA is ligated directly to its targets to form a single hybrid molecule. Each library produces hybrid reads in addition to individual reads. AGO-bound ribonucleic molecules were ligated with RA3 primer containing an IRDye 800CW, separated on a NuPAGE Bis-Tris gel, and transferred to nitrocellulose. Blots show AGO-RNA complexes detected on nitrocellulose membrane. (C) 32P Northern blot of AGO-bound RNAs in HCT116 cells probed for miR-16 and miR-320a. (D) Non-hybrid miRNA reads from qCLASH libraries were quantitated with miR-Deep2. Normalized counts from qCLASH libraries that used the same AGO antibody were averaged, showing that both 2A8 and 4F9 pull down similar amounts of each miRNA. The correlation coefficient (r) is depicted on the graph.
Fig 2.
Analysis of AGO-qCLASH in HCT116 colorectal cancer cells.
Venn diagram depicting the overlap of miR-320a target genes in qCLASH and CLASH for (A) WT cells and (B) DROSHA KO cells. (C) Normalized read count of miRNAs in HCT116 and HCT116 DROSHA KO cells. miRNA counts were determined with miR-Deep2 and normalized to total reads. TSS-miRNAs are labeled in magenta, mirtrons are labeled in green, miR-snaR is labeled in purple, and all other miRNAs are labeled in black. (D) Correlation plot of miRNA abundance and hybrid counts for each miRNA. miRNA counts were determined using miR-Deep2 and normalized to total number of reads. The number of hybrids for each miRNA was determined and plotted against miRNA counts. A correlation coefficient (r) is provided. (E) miRNA target sites are found predominately in the CDS and 3′UTR. The location of each transcript from the mRNA portion of the hybrid was determined using data from Ensemble Biomart. Each hybrid was assigned a location based on where it was located in relation to the beginning and end of the reference CDS. Transcripts with IDs not found in the database were excluded from the analysis. (F) The proportion of hybrids with seed binding was less than non-canonical pairings. Seed binding was determined using viennad files, which predicts folding between the miRNA and its target. Individual miRNAs display distinct distributions of seed bindings.
Table 1.
Summary of AGO-qCLASH in HCT116 cells.
The table depicts the total number of miRNA/mRNA reads in each qCLASH sample. miR-320a/mRNA reads are broken down by replicate and summed for each sample group. Percentage depicts the fraction of miR-320a hybrids in total miRNA/mRNA hybrids. BR = Biological replicate.
Fig 3.
Identification of pathways targeted by miR-320a in HCT116 cells.
(A) Top ten pathways targeted by miR-320a in WT qCLASH. The corresponding -log (p-value) for each pathway in DROSHA KO qCLASH is depicted right of each bar. The yellow line defines significance threshold for IPA. (B) The cumulative distribution function of fold change in gene expression for qCLASH-identified miR-320a targets (in ≥ 3 WT replicates) following miR-320a mimic transfection in HCT116 DICER KO cells. Green line: all qCLASH identified targets; Blue line: qCLASH-identified targets containing only seed matches; Yellow line: qCLASH-identified targets containing only non-seed matches; Red line: miR-qCLASH-identified targets that contain both seed and non-seed matches; Gray line: conserved miR-320a targets predicted by TargetScan; Black line: all transcripts. P-values were determined using Kolmogorov Smirnov tests between colored subsets and control (black). (C) The cumulative distribution function of fold change in gene expression for qCLASH-identified miR-320a targets (in ≥ 4 WT replicates, blue line) following miR-320a antagomir transfection in HCT116 DROSHA KO cells. (D) Top 10 pathways enriched in differentially expressed genes (p ≤ 0.05) from (B) were determined by IPA. (E) High-confidence targets from qCLASH were validated using RT-qPCR. HCT116 DICER KO cells were transfected with miR-320a mimic or non-specific control. Total RNA was prepared and gene expression was measured with RT-qPCR using the comparative Ct method and depicted as fold changes. Data are shown as the average of three biological replicates with a standard deviation. ns = P > 0.05, * = P ≤ 0.05, ** = P ≤ 0.01, **** = P ≤ 0.0001.
Fig 4.
miR-320a targets CANX in colorectal cancer cells to activate UPR.
(A) CANX expression is reduced in miR-320a mimic transfected DICER KO cells, measured by RNA-seq. Gene counts were generated from Stringtie output and normalized using Deseq2. Individual samples were plotted along with mean and standard deviation, for three biological replicates. * = P ≤ 0.05. Western blots of CANX in miR-320a mimic (B) and antagomir (C) transfected HCT116 cells. GAPDH was used as a loading control. Quantitation of CANX protein expression is depicted in bar graphs as individual points with standard deviation for three biological replicates. * = P ≤ 0.05, *** = P ≤ 0.001. (D) Predicted base-pairing of miR-320a and miR-HSUR4-3p with CANX 3′ UTR; miR-320a seed match sequence on CANX is highlighted in orange; miR-HSUR4-3p seed match sequence is highlighted in blue. (E) Dual-luciferase reporter containing WT CANX 3′ UTR miR-320a binding sites, a mutation in site 1 (site 1 mut), a mutation in site 2 (site 2 mut), or double mutant was co-transfected with a vector expressing miR-320a, miR-HSUR4-3p, or an empty vector in 293T cells. All luciferase assays were performed in technical and biological triplicates. ns = P > 0.05, * = P ≤ 0.05, ** = P ≤ 0.01.
Fig 5.
miR-320a activates ATF4 during ER stress.
(A) RT-qPCR quantification of spliced(s)-XBP1, BiP, and CHOP mRNA in HCT116 cells transfected with miR-320a mimic. Data are shown as the average of three biological replicates with a standard deviation and individual data points. All three mRNAs were normalized to HPRT. Experiments were performed in technical triplicate. (Student’s t-test): ns = P > 0.05, * = P ≤ 0.05, ** = P ≤ 0.01. (B) Western blots of ATF4 in miR-320a mimic-transfected WT cells with thapsigargin (Tg), tunicamycin (Tm) or histidinol (HisOH) treatment. (C) Western blots of ATF4 in miR-320a antagomir-transfected HCT116 with Tg treatment. Western blots of Atf4 in miR-320a mimic-transfected MEF WT (D) and Perk -/- cells (E) with Tg treatment. In all western blots, GAPDH was used as a loading control. Quantitation of three biological replicates with mean and standard deviation is shown with individual data points. (Student’s t-test): ns = P > 0.05, * = P ≤ 0.05, **** = P ≤ 0.0001.