Terminal Uridylyltransferases TUT4/7 Regulate microRNA and mRNA Homeostasis

Abstract

The terminal nucleotidyltransferases TUT4 and TUT7 (TUT4/7) regulate miRNA and mRNA stability by 3' end uridylation. In humans, TUT4/7 polyuridylates both mRNA and pre-miRNA, leading to degradation by the U-specific exonuclease DIS3L2. We investigate the role of uridylation-dependent decay in maintaining the transcriptome by transcriptionally profiling TUT4/7 deleted cells. We found that while the disruption of TUT4/7 expression increases the abundance of a variety of miRNAs, the let-7 family of miRNAs is the most impacted. Eight let-7 family miRNAs were increased in abundance in TUT4/7 deleted cells, and many let-7 mRNA targets are decreased in abundance. The mRNAs with increased abundance in the deletion strain are potential direct targets of TUT4/7, with transcripts coding for proteins involved in cellular stress response, rRNA processing, ribonucleoprotein complex biogenesis, cell-cell signaling, and regulation of metabolic processes most affected in the TUT4/7 knockout cells. We found that TUT4/7 indirectly control oncogenic signaling via the miRNA let-7a, which regulates AKT phosphorylation status. Finally, we find that, similar to fission yeast, the disruption of uridylation-dependent decay leads to major rearrangements of the transcriptome and reduces cell proliferation and adhesion.

Keywords: RNA degradation; miRNA/mRNA network; microRNA homeostasis; oncogenic signaling; transcriptome; uridylation dependent decay.

Figure 1

TUT4/7 regulate miRNA and gene expression levels. (A) miRNAs are regulated by TUT4/7 in multiple pathways: (i) Pre-miRNAs can be polyuridylated by TUT4/7 in the presence of LIN28A, a signal for degradation by DIS3L2. Decreased miRNA levels due to DIS3L2-mediated degradation increase target mRNA levels. (ii) Group I pre-miRNAs contain a 2-nucleotide 3′ overhang and can be processed by DICER to form the mature miRNA. This process occurs independently of TUT4/7 regulation of miRNAs. (iii) Group II pre-miRNAs contain a 1-nucleotide 3′ overhang and must be extended by one nucleotide at the 3′ end to form the necessary 2-nucleotide 3′ overhang for DICER processing. Mature miRNAs are processed into a guide (usually 5p) strand, which is incorporated into the RNA-induced silencing complex (RISC), and a passenger (usually 3p) strand, which is degraded. Mature miRNAs reduce expression of target mRNAs. (B) mRNAs can be 3′ polyuridylated by TUT4/7 and degraded by DIS3L2, decreasing gene expression.

Figure 2

MicroRNAs are differentially expressed in HEK 293T cells lacking TUT4/7. (A) Volcano plot of miRNA expression plotting log2 fold change, in ΔTUT4/7 relative to wild-type HEK 293T cells, versus false discovery rate (FDR). Significantly (FDR ≥ 1, log2FC ≥ 1 or ≤ −1) changed miRNAs are indicated in red (upregulated, 115 miRNAs),blue (downregulated, 91 miRNAs) or not significantly changed miRNAs in black (622 miRNAs). (B) Bar graph representing the results of a microRNA family set analysis using TAM2.0, showing numerous miRNA families are enriched in significantly up- (orange) and downregulated (blue) miRNAs. Enrichment is expressed as log (p value) of the significance for miRNA enrichment. (C) RNA sequencing (orange) and RT-qPCR (orange and grey hashmarks) analysis reveal the different regulatory patterns between group I and group II let-7 miRNAs in ΔTUT4/7 cells: group I let-7 miRNAs have relatively small changes in expression in ΔTUT4/7 cells, while most group II let-7 miRNAs are drastically upregulated. Let-7a is encoded at multiple genomic locations as both group I and group II pre-miRNAs. All fold change values are shown relative to wild-type miRNA expression levels. Significance was calculated by Partek gene-specific analysis (GSA) (RNA-seq) or Student’s t-test (RT-qPCR) and is indicated by asterisks, where * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Figure 3

Messenger RNAs are differentially regulated in HEK 293T cells lacking TUT4/7. (A) Volcano plot of mRNA log2 expression fold change (FC) in ΔTUT4/7, relative to wild-type HEK 293T cells, versus false discovery rate (FDR). Significantly (FDR ≥ 1, log2FC ≥ 1 or ≤ −1) changed mRNAs are indicated in red (upregulated, 541 mRNAs) or blue (downregulated, 1573 miRNAs). The mRNAs not significantly changed in expression are depicted in black. (B) Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) analysis reveals a physical subnetwork of mRNAs that are at least 2-fold upregulated in ΔTUT4/7 cells. GO analysis of the STRING network shows enrichment in the following: ribonucleoprotein complex biogenesis (red, strength 0.53), regulation of metabolic process (yellow, strength 0.12), cell–cell signaling (green, strength 0.27), and RNA processing (blue, strength 0.29). Differentially expressed mRNAs not included in these GO terms are indicated in white; those with no interactions to others are not included in the network. Confidence strength of data support was calculated by log10 (observed/expected) and is indicated by line thickness, with thicker lines indicating stronger supporting data. (C) RNA sequencing (orange) and RT-qPCR (orange and grey hashmarks) of mRNAs involved in heat shock response in ΔTUT4/7 cells. Fold change is calculated relative to wild-type HEK 293T cells. (D) Western blotting and (E) quantification of heat shock response protein levels in WT (grey) and ΔTUT4/7 (orange) HEK 293T cells. Significance was calculated by Partek gene specific analysis (GSA) (RNA-seq) or Student’s t-test (RT-qPCR, Western blot) where ns = not significant; * p < 0.05,**** p < 0.0001.

Figure 4

Messenger RNAs downregulated in ΔTUT4/7 cells are implicated in changes to cellular phenotype. (A) Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) analysis reveals a physical subnetwork of mRNAs that are at least 2-fold downregulated in ΔTUT4/7 cells with a stringent false discovery rate to reduce nodes (FDR < 1⁻¹⁰). Gene Ontology (GO) analysis of the STRING network shows enrichment of cellular processes (red, strength 0.06), circulatory system development (blue, strength 0.42), regulation of developmental processes (light green, strength 0.26), anatomical structure morphogenesis (yellow, strength 0.28), and tube development (dark green, strength 0.41). Differentially expressed mRNAs not included in these GO terms are indicated in white; those with no interactions to others are not included in the network. Confidence strength of data support is calculated by log10 (observed/expected) and indicated by line thickness, with thicker lines indicating stronger supporting data. (B,C) Representative images of (B) wild-type and (C) ΔTUT4/7 HEK 293T cells 72 h after plating at 5 × 10⁴ cells per mL show multicellular aggregates in ΔTUT4/7 cells (black arrows). (D) CCK-8 cell proliferation assay of wild-type (black circles) versus ΔTUT4/7 (orange squares) HEK 293T cells. (E) Quantification of a crystal violet cell adhesion assay of wild-type (black circles) versus ΔTUT4/7 (orange squares) HEK 293T cells. Significance was calculated by 2-way ANOVA in GraphPad Prism and is indicated by asterisks, where ns = not significant; * p < 0.05; ** p < 0.01; *** p < 0.001.

Figure 5

The miRNA–mRNA interaction network reveals target distribution among 1933 target mRNAs (pink circles) of the upregulated let-7 miRNAs (blue squares), 94 are downregulated at least two-fold (highlighted in bigger red circles) in ΔTUT4/7 HEK 293T cells. The miRNA–target interaction is based on experimentally validated information obtained from miRTarbase v8.0. The diagram was generated using miRNet 2.0.

Figure 6

Messenger RNA and protein abundance are not directly related in the TUT4/7 deletion. (A) Western blots and (B) quantification of AKT isozymes in wild-type (black, circles) and ΔTUT4/7 (orange, squares) HEK 293T cells. (C) Artificial increase in let-7a levels by transfection was shown previously to increase T308 phosphorylation and reduce S473 phosphorylation of AKT1 in response to epidermal growth factor stimulation. (D) Western blots and (E) quantification of pan-AKT phosphorylation in wild-type (black, circles) and ΔTUT4/7 (orange, squares) HEK 293T cells. (F) Deletion of TUT4/7 alters response to EGF stimulation, increasing phosphorylation of AKT at T308 and reducing the response at S473. AKT1 specifically is decreased in ΔTUT4/7 cells. Blots were cropped to remove empty lanes (A) or molecular weight guides (D) for clarity. Asterisks indicate significance in Students t-test or ANOVA in GraphPad Prism where ns = not significant; * p < 0.05.