A dual-activity topoisomerase complex regulates mRNA translation and turnover

Abstract

Topoisomerase 3β (TOP3B) and TDRD3 form a dual-activity topoisomerase complex that interacts with FMRP and can change the topology of both DNA and RNA. Here, we investigated the post-transcriptional influence of TOP3B and associated proteins on mRNA translation and turnover. First, we discovered that in human HCT116 colon cancer cells, knock-out (KO) of TOP3B had similar effects on mRNA turnover and translation as did TDRD3-KO, while FMRP-KO resulted in rather distinct effects, indicating that TOP3B had stronger coordination with TDRD3 than FMRP in mRNA regulation. Second, we identified TOP3B-bound mRNAs in HCT116 cells; we found that while TOP3B did not directly influence the stability or translation of most TOP3B target mRNAs, it stabilized a subset of target mRNAs but had a more complex effect on translation-enhancing for some mRNAs whereas reducing for others. Interestingly, a point mutation that specifically disrupted TOP3B catalytic activity only partially recapitulated the effects of TOP3B-KO on mRNA stability and translation, suggesting that the impact of TOP3B on target mRNAs is partly linked to its ability to change topology of mRNAs. Collectively, our data suggest that TOP3B-TDRD3 can regulate mRNA translation and turnover by mechanisms that are dependent and independent of topoisomerase activity.

Figure 1.

Ribo-seq and RNA-seq reveal diverse changes of gene expression in TOP3B-KO HCT116 cells. (A) A schematic diagram displaying the experimental design procedure. The details are described in the Results. (B) Volcano plots showing the numbers of down-regulated (left side) or upregulated (right side) differentially expressed genes (DEGs) in TOP3B-KO by RNA-seq (blue) or Ribo-seq (orange) from five independent experiments. The DEGs that showed statistically significant difference between KO versus WT were marked by colors, and were identified using DESeq2 (Fold change > 1.5, adjust P-values < 0.1). Fold changes and adjust P-values were log-transformed. The numbers of decreased (down) or increased (up) DEGs are indicated in the graphs. The DEGs that did not show statistically significant difference were marked by gray color. (C) A heatmap shows expression changes (in fold change) between TOP3B-KO versus WT for the DEGs identified by Ribo-seq. The DEGs with decreased or increased signals were marked by blue and red, respectively. The decreased or increased DEGs were each divided into four groups based on their RPF, RNA-seq and PRO-seq levels as described in the main text. Groups I and II represent DEGs altered in mRNA translation and turnover (stability) but not transcription; whereas Groups III and IV represent DEGs altered in transcription. Several representative genes altered at post-transcriptional levels were marked on the right.

Figure 2.

TOP3B regulates translation and turnover of specific mRNAs. (A, B) BedGraphs (A), and their quantifications (B), show four representative genes with altered RPF levels in TOP3B-KO cells. eCLIP-seq results show the binding of TOP3B to three of them. The RNA-seq and PRO-seq were included for comparison. The RNA-seq level for FAT1 was significantly altered in TOP3B-KO cells, whereas its PRO-seq level remained unchanged, suggesting that turnover of this mRNAs was altered. GAPDH was included as a control. The black arrows above the bedGraphs mark the TOP3B eCLIP peaks that were higher in WT and lower in TOP3B-KO cells (a negative control), whereas those below mark the transcriptional direction. Blue and red arrowheads (below) mark the positions of translation start and stop codons separately. Blue and red arrows (middle) mark the decrease or increase of the expression level separately. The data in (B) were shown in normalized counts generated by DESeq2. * = adjusted P-value < 0.1, ** = adjust P-value < 0.05, *** = adjust P-value < 0.01.

Figure 3.

TOP3B co-regulates more genes with TDRD3 than with FMRP. (A) Schematic representation of the TOP3B–TDRD3-FMRP complex. (B, C) Volcano plots showing the numbers of down-regulated or upregulated DEGs in TDRD3-KO and FMR1-KO by RNA-seq (blue) or Ribo-seq (orange) from three independent experiments. Description of Volcano plots are described in Figure 1 Legend. (D, E) Heatmaps showing the concomitantly decreased (blue color) or increased (red color) expression changes of the DEGs in TOP3B-KO versus those of TDRD3-KO and FMR1-KO cells by RNA-seq (D) or Ribo-seq (E). The percentages below the maps were calculated by artificially setting the decreased or increased DEGs of TOP3B-KO cells as 100% (column 1). The percentages of the DEGs of TDRD3-KO or FMR1-KO cells that were altered in the same directions were shown in a table below the figure. The cutoff threshold for the increased or decreased DEGs is 1.5-fold. Notably, a stronger co-clustering was observed between TOP3B-KO and TDRD3-KO than that between TOP3B-KO and FMR1-KO (column 4 versus 7). (F) Graphs to compare the percentages of overlapping DEGs in the same or opposite direction of alteration vs. those randomly selected genes between TOP3B-KO and TDRD3-KO cells. The DEGs were identified by either RNA-seq (blue) or Ribo-seq (orange). The percentages were relative to the DEGs numbers of TOP3B-KO cells. The randomly selected genes were expression level matched, and the numbers were identical to that of the decreased or increased DEGs of TDRD3-KO cells. The percentages of these genes that were decreased or increased in RNA-seq or Ribo-seq of TDRD3-KO cells were then calculated. Blue arrows represented reduced, whereas red arrows represented increased DEGs. Arrows in the same direction depicted DEGs that were altered in the same direction in TOP3B-KO and TDRD3-KO cells. (G) The DEGs between TOP3B-KO and FMR1-KO were analyzed using the same method as (F). Note the scales on Y-axis are smaller in (G) than (F).

Figure 4.

TOP3B and TDRD3 co-regulate translation and turnover of a group of mRNAs. (A, B) BedGraphs of sequencing read distributions (A), and bar graphs of quantification of these reads (B), show the Ribo-seq, mRNA and PRO-seq signals for five representative genes in TDRD3-KO and FMR1-KO cells. The RNA-seq levels for CHD8, SMC3, EML3 and FAT1 were altered in TDRD3-KO cells, whereas their PRO-seq levels remain unchanged, suggesting that turnover of these mRNAs was altered. Black arrows below mark the transcription direction. Blue and red arrowheads (below) mark the positions of translation start and stop codons separately. Blue and red arrows (middle) mark the decrease or increase of the expression level separately. The alteration for these mRNAs in TOP3B-KO and Y336F-KI cells are described in Figures 2A and 8A. Normalized counts were generated by DESeq2. *** = adjust P-value < 0.01.

Figure 5.

TOP3B preferentially binds coding region of long mRNAs and increases their stability. (A) Schematic overview of the major steps of eCLIP-seq (41). (B) A pie chart displaying TOP3B eCLIP-seq read density (tags/kb) distribution in CDS (coding sequence), 5’UTR, 3’UTR and introns. Reads from WT immunoprecipitation group were analyzed using RSeQC. (C) Box and Whisker plot showing longer average lengths of mRNAs bound by TOP3B than those that are unbound. The TOP3B-bound mRNAs were identified by eCLIP in HCT116 or HITS-CLIP in HeLa cells. The mRNAs bound by FMRP identified by HITS-CLIP in mouse brains were also shown (3,58). The mRNA lengths were log-transformed. (D) Scatter plots showing the strong correlations of RNA, RPFs or PRO-seq levels of TOP3B eCLIP targets between WT and TOP3B-KO cells. Notably, there are more mRNAs with reduced RNA-seq levels in TOP3B-KO HCT116 cells (left panel), whereas the numbers of mRNA with reduced or increased PRO-seq signals are comparable in the same cells (right panel), indicating increased turnover of TOP3B-bound mRNAs in the absence of TOP3B. (E) Box-Whisker plots showing that TOP3B eCLIP target mRNAs exhibit overall reduced levels of mature transcripts (left), but unchanged levels of nascent transcripts (right), in TOP3B-KO cells compared with WT cells. As a control, the randomly selected unbound mRNAs (also expression level matched) did not display this trend. These data suggest that TOP3B binding stabilizes its target mRNAs by reducing turnovers. ***P-value < 0.01; n.s., no significant.

Figure 6.

TOP3B binding preferentially reduces mRNA turnover but affects translation in either directions. (A, B) Volcano plots showing the expression changes of TOP3B eCLIP targets by RNA-seq (A) or Ribo-seq (B). The target mRNAs were colored in blue or orange, whereas the non-target mRNAs were in gray. The target mRNAs exhibiting significant difference between KO vs. WT cells were shown in darker blue or orange colors. The description of Volcano plots were in Figure 1 Legend. (C, D) Heatmaps showing the expression changes of the TOP3B-bound mRNAs determined by RNA-seq (C) or Ribo-seq (D), in different KO and KI mutant cells indicated on the top. The percentages below the maps were calculated by artificially setting the decreased or increased DEGs of TOP3B-KO cells as 100% (column 1 of each graph). The percentages of these DEGs altered in the same directions in other KO or KI mutant cells were shown in a table below the figure. The DEGs identified by different Seq methods from each cell type were included in the analysis. The cutoff threshold for the increased or decreased DEGs is 1.5-fold. Notably, the TOP3B targets with altered mRNA abundance were nearly all deceased in TOP3B-KO cells by RNA-seq (C, lane 1), whereas only 9% of them were decreased by PRO-seq (lane 3), indicating that majority (91%) of these mRNAs have increased turnover. Moreover, the TOP3B targets with altered RPF levels were either increased or decreased (D, lane 1), indicating that TOP3B binding can affect translation in either directions.

Figure 7.

TOP3B regulates mRNAs in its topoisomerase activity dependent and independent manners. (A) Volcano plots showing the number of DEGs in Y336F-KI cells by RNA-seq (up) or Ribo-seq (down). The description of Volcano plots is in Figure 1B Legend. (B) Graphs to compare the percentages of overlapped DEGs in the same or opposite directions of alteration between Y336F-KI and TOP3B-KO groups vs. those randomly selected genes. The percentages were relative to the DEGs numbers of TOP3B-KO group. The randomly selected genes were expression level matched, and the numbers were identical to that of the decreased or increased DEGs of Y336F-KI cells. Blue arrows represent reduced, whereas red arrows represent increased DEGs. Arrows in the same direction depicted DEGs that were altered in the same direction in TOP3B-KO and Y336F-KI cells. (C, D) Heatmaps display the DEGs in TOP3B-KO which were overlapped with those of Y336F cells by RNA-seq (C) or Ribo-seq (D). The DEGs with decreased or increased signals (fold changes) were marked by blue and red, respectively. The overlapped DEGs that were altered in the same directions were marked as “+”, whereas the others were marked as “–”. The percentages of the overlapped or non-overlapped DEGs were shown on the right. (E) Venn diagrams showing the overlapped DEGs between TOP3B-KO and Y336F-KI cells.

Figure 8.

TOP3B requires its topoisomerase activity to regulate translation and turnover of specific mRNAs. (A) BedGraphs of sequencing read distributions, and bar graphs of quantification of these reads (B), show the Ribo-seq, RNA-seq and PRO-seq signals for five representative genes. The RNA-seq levels for CHD8, SMC3, EML3 and FAT1 mRNAs were altered in Y336F-KI cells, whereas their PRO-seq levels remain unchanged, suggesting that turnover of these mRNAs was altered. The alteration of these genes in TOP3B-KO cells were described in Figure 2A. The bar graphs show the normalized counts from three biological replicates. *** = adjust P-value < 0.01. (C, D) Heatmaps showing how the concomitantly decreased (blue color) or increased (red color) DEGs in both TOP3B-KO and TDRD3-KO cells were overlapped with the DEGs in TOP3B-Y336F and FMR1-KO cells by RNA-seq (C) or Ribo-seq (D). The percentages below the maps were calculated by artificially setting the decreased or increased DEGs of both TOP3B-KO and TDRD3-KO cells as 100% (column 1 and 4 of each graph). The percentages of the DEGs of TOP3B-Y336F or FMR1-KO cells that were altered in the same directions were shown in a table below the figure. The DEGs identified by different Seq methods from each cell type were included in the analysis. The cutoff threshold for the increased or decreased DEGs is 1.5-fold. Notably, a stronger co-clustering was observed between TOP3B-KO/TDRD3-KO and TOP3B-Y336F than that between TOP3B-KO/TDRD3-KO and FMR1-KO (column 7 versus 10). In addition, a stronger co-clustering was also detected between the levels of RPF and RNA than with PRO-seq for each cell type.

Figure 9.

The deficiency of TOP3B reduces the translation of CHD8 mRNA resulting in a reduction of CHD8 mRNA stability. (A) Graphs from polysome profiling analysis show the reduced translation of CHD8 mRNA in TOP3B-KO cells, as evidence by its decreased level in the heavy polysome fractions, and increased level in light polysome fractions. The relative distributions (%) of the mRNAs on the sucrose gradients were quantified by RT-qPCR analysis. TOP3B and GAPDH were used as controls. The translation of TOP3B mRNA was significantly reduced in TOP3B-KO1 cells caused by pre-mature stop codon induced by CRISPR-Cas9 editing. No obvious difference of the translation was observed for GAPDH mRNA. The results were reproducible, but only one representative result was shown here. (B) Immunoblotting images and quantification show that CHD8 protein levels were reduced in TOP3B and TDRD3 mutant cells, but not in FMR1-KO cells, as indicated on top. (C) The nascent mRNA levels of representative genes were detected by RT-qPCR. The results showed that there was no significant change for TOP3B gene, slightly increased for CHD8 and significantly reduced for ANXA10 gene, which were consistent with the PRO-seq results (Supplementary Table S1). (D, E) RT-qPCR results showing the mRNA levels of the representative genes. Cycloheximide (CHX) treatment (100μg/ml) for 3 hours increased TOP3B and CHD8 mRNA levels in TOP3B-KO cells compared with non-treatment cells (D). CHX treatment also increased CHD8 mRNA levels in TDRD3-KO and Y336F-KI cells (E). These data indicate that the reduced mRNA levels of TOP3B and CHD8 were translation related. For the bar graphs, all the values are normalized to WT and log-transformed before P-value calculation *P-value < 0.05, **P-value < 0.01.

Figure 10.

Models showing the roles of the TOP3B–TDRD3 complex in mRNA translation and turnover. (A) Models to explain how TOP3B–TDRD3 can regulate mRNA translation in topoisomerase activity dependent (left panel) and independent manners (right panel). The left panel shows three models. One, the TOP3B–TDRD3 complex may resolve (decatenate) two tangled mRNAs to promote their translation and stability. This model is supported by in vitro assays showing TOP3B can catalyze catenation of RNA circles (3,8). Two, an mRNA may become topologically constrained by circularization during translation. A supercoil or knot-like structure may be formed and required an RNA topoisomerase to release (3). It is possible that TOP3B may induce formation of the above topological structures to repress translation. Three, The TOP3B–TDRD3 complex may promote translation initiation by resolving topological stress at 5’UTR regions. The right panel shows two models of enzymatic activity independent mechanisms. The first model suggests the TOP3B–TDRD3 complex may recruit EJC complex to its target mRNAs to enhance their translation and reduce their degradation by nucleases (2). The second model depicts that TOP3B–TDRD3 may regulate translation of its targets by recruiting FMRP (2,3,9). (B) A cartoon shows two models to explain how TOP3B regulates translation of mRNAs that show continuous reduction of RPF levels across the entire CDS region. This RPF pattern (top) has been observed in all representative differentially-expressed mRNAs in the main figures (Figure 2A). Both models predict that inactivation of TOP3B–TDRD3 induces topological stress in mRNAs (marked by a cross). The first model hypothesizes that the stress inhibits initiation, whereas the second model suggests that it inhibits elongation. The second model fits mRNAs (such as CHD8) that are subject to translation-associated mRNA decay(46), whereas the first model fits other mRNAs (such as SMC3) that show no reduction of their levels. See Supplementary Figure S11 for RPF patterns of CHD8 and SMC3 mRNAs on codon-by-codon beGraphs. The details are described in the main text. The nucleases are depicted by yellow.