Creating TOP3B-TDRD3 KO and KI cell lines to study their function
To study the function of the TOP3B-TDRD3 complex, we generated isogenic HCT116 cell lines that are individually inactivated for TOP3B (TOP3B-KO); TDRD3 (TDRD3-KO); the catalytic activity of TOP3B (Y336F-KI); or their interacting protein FMRP (FMR1-KO), using the CRISPR-Cas9 technology (Fig. 1a-d & Fig. S1a-f) (19). We then utilized these cell lines to identify TOP3B-TDRD3 DNA and RNA targets by ChIP-seq and eCLIP-seq analyses, respectively. We further analyzed the effect of TOP3B-TDRD3 mutation on transcription by pol II ChIP-seq and PRO-seq; mature mRNA levels by RNA-seq; and translation by Ribo-seq (Fig. 1a) (20-22). The HCT116 cell line has earlier been extensively used for comparable transcription and translation studies, enabling us to compare our findings for TOP3B with those for other topoisomerases. Genomic DNA sequencing revealed the occurrence of cleavage-directed frame-shift mutations in various KO clones (Fig. S1b & S1e), as well as the targeted homozygous Y336F-substitutions in the KI clone (Fig. 1c). Western blotting confirmed the absence of each target protein in their respective KO cells (Fig. 1b) and the presence of the Y336F mutant protein in TOP3B-KI cells (Fig. 1d). The level of Y336F was about 50% of that of WT cells, suggesting that the catalytic reaction may help to stabilize TOP3B protein. We noticed that the levels of TOP3B and TDRD3 were lower in KO cells of each other (~70%), but not in FMR1-KO cells (Fig. 1b), which is consistent with earlier observations that TOP3B-TDRD3 depends on each other for stability (7).
We generated an additional TOP3B-KO clone (KO2) using a different guide RNA and found that it expresses a mutant protein carrying a partial deletion of the Toprim domain (Fig. S1c-d). Because this highly conserved domain is critical for topoisomerase activity (23), the mutant protein is expected to show loss of function and was used as an internal control in our translation studies (see below).
TOP3B-TDRD3 complex regulates transcription of a subset of genes.
We first studied whether genome-wide transcription is altered in cells with a KO or KI mutant TOP3B-TDRD3 complex by pol II ChIP-seq and PRO-seq analyses. The former method detects both transcribing and stalled pol II binding genome-wide, whereas the latter measures the levels of nascent transcripts produced by an active polymerase, and thus can more specifically reflect transcription (21). Scatter plot analysis revealed that the PRO-seq and pol II signals at transcription start sites (TSS) of all genes in TOP3B-KO, TDRD3-KO, and TOP3B-Y336F-KI cells strongly correlate with those of WT cells, with a correlation coefficient > 0.9 (Fig. 1e; Fig. S2a; Fig. S2c), indicating that inactivation of TOP3B-TDRD3 complex leaves global transcription largely unperturbed in basal conditions.
In fact, calculation of PRO-seq signals at TSS and exons revealed that the majority of genes remained unchanged (>66%, 1.2-fold threshold), whereas a smaller percentage of genes showed consistently decreased or increased in two independent biological replicates (Fig. 1f; Fig. S2b; Table S1). A Venn graph analyses identified about 244 genes consistently decreased in both TOP3B-KO and TDRD3-KO cells; 224 genes that are decreased in both TOP3B-KO and Y336F-KI cells; 132 genes that are decreased in all three cell lines (Fig. 1g & Table S1). We infer that these 132 genes are likely regulated by the TOP3B-TDRD3 complex in a topoisomerase activity dependent manner, either directly or indirectly.
We noticed that the number of upregulated genes in TOP3B-KO and TDRD3-KO cell lines is about 3 times greater than that of the down-regulated genes (Fig. 1f). This was somewhat unexpected, because topoisomerase is known to enhance but not repress transcription. One explanation could be that the upregulated (and some downregulated) genes are indirectly regulated by TOP3B-TDRD3.
A gene ontology (GO) analysis of the downregulated genes showed enrichment of functional terms related to regulation of transcription (Fig. 1h), consistent with the notion that the upregulated genes might result from the alleviation of secondary effects of TOP3B-TDRD3 inactivation. The altered transcription of several genes in TOP3B-TDRD3 KO and KI cells by PRO-seq was verified by bedGraph analysis and subsequently by RNA-seq analysis of polyA-selected RNA (Fig. 1i; Fig. S2d-e). The altered PRO-seq signals at TSS suggest that the initiation of transcription is regulated by the TOP3B-TDRD3 complex in a topoisomerase dependent manner. One of them is FGF2, which encodes a fibroblast growth factor that is reduced in the postmortem brains of depressed patients (24). Another example is APOBEC3H, which encodes a protein functioning importantly in innate anti-viral immunity (25). Dysregulation of these genes may contribute to the autism observed in individuals carrying a TOP3B mutation, and to the autoimmune phenotype observed in TOP3B-KO mice (26).
TOP3B-TDRD3 preferentially binds TES regions of highly active genes
We identified TOP3B-TDRD3-bound genes in HCT116 cells by ChIP-seq, using two different anti-TDRD3 antibodies. Mock ChIP-seq in TDRD3-KO cells was included as a negative control, which is critical because TOP3B-TDRD3 preferentially binds highly transcribed genes (6) that are “hotspots” in ChIP-seq (27). We calculated the difference in TDRD3 ChIP signals at TSS, exons, introns and transcription end sites (TES) between WT and TDRD3-KO cells from three replicates and identified a total of 55 genes that have >1.2-fold difference at any of the four regions (Table S2). We considered these as TOP3B-TDRD3-bound genes. Interestingly, the number of genes with TOP3B-TDRD3 binding at TES (23) is greater than the number with binding at TSS (20) or other regions (15). BedGraph visualization confirmed the presence of strong TDRD3 peaks in several representative genes from WT but weaker signals from TDRD3-KO cells. These include EGR1 and AREG, in which TDRD3 binds at TES; and ACSL5, in which TDRD3 binds at TSS (Fig. 2a). The TDRD3 peaks at either TES or TSS always colocalized with the position of pol II, consistent with our previous findings that TOP3B-TDRD3 associates with pol II during the entire transcription process in mouse brains (6). As a negative control, no obvious difference between WT and TDRD3-KO cells was detected for a TOP3B-TDRD3 unbound gene, GAPDH (Fig. 2a).
We found that the average PRO-seq and Pol II signals of TOP3B-TDRD3 bound genes were significantly higher compared to the average signals over all genes (Fig. 2b), indicating that the complex preferentially binds highly transcribed genes, in accord with our previous findings in mouse brains under neuronal activation (6).
TOP3B topoisomerase activity facilitates transcription of some of its target genes
Analysis of the TOP3B-TDRD3 bound genes revealed that their PRO-seq and pol II signals at TSS, exons and TES in TOP3B-KO cells strongly correlate with those of the corresponding regions in WT cells (Fig. 2c). The signals for only a subset of genes showed consistent reduction. The results resemble our previous mouse data that many TOP3B-TDRD3-bound genes exhibit no obvious decrease in basal transcription.
Among the altered genes, EGR1, AREG and ACSL5 showed reduced nascent transcript levels at TSS, exons and TES, ranging from -20% to -80%, in both TOP3B-KO and TDRD3-KO cells, by bedGraph visualization (Fig. 2a) and quantification from two independent replicates of PRO-seq (Fig. 2d), suggesting that the TOP3B-TDRD3 complex binds and stimulates transcription of these genes. The reduced transcription of these genes in TOP3B-KO cells was confirmed by pol II ChIP-seq (Fig. 2a, e), and RT-qPCR (Fig. 2f) analyses. As a control, no obvious difference was observed for GAPDH (Fig. 2a, d, e). The strongest reduction was observed in the TES of AREG (80%), greater than that at TSS or exons of the same gene, suggesting that the complex enhances transcription elongation of this gene.
The reduction of PRO-seq signal in TOP3B-Y336F mutant cells was comparable to those in TOP3B-KO cells at TSS and exons, but not at TES for the above genes (Fig. 2d). Analysis of EGR1 in this cell line showed that its pol II levels displayed significant reduction at exons (p<0.05), and a strong trend of reduction at TES (p=0.07) (Fig. 2e); and EGR1 mRNA levels also exhibited a strong trend for reduction (p=0.08) (Fig. 2f), suggesting that TOP3B catalytic activity contributes to transcription of this gene. In contrast, for AREG, its pol II and RNA-seq signals exhibit no obvious decrease (or perhaps a small increase), indicating that TOP3B catalytic activity is dispensable for transcription of this gene (Fig. 2e-f). For ACSL5, its pol II and RT-qPCR levels showed reduction, but not statistically significant. Taken together, these data suggest that the TOP3B-TDRD3 complex can coordinately bind and regulate transcription of specific genes using topoisomerase activity dependent and independent mechanisms.
TOP3B inactivation disrupts translation and mRNA levels for a small number of genes
We examined mRNA translation in HCT116 cells harboring mutations of either TOP3B or TDRD3 by Ribosome profiling (Ribo-seq), which monitors translation of mRNAs genome-wide by quantifying ribosome-protected RNA fragments (RPFs) (20). We also analyzed the levels of mRNAs genome-wide in the same experiment by RNA-seq. As a proof of validity of our assays, the levels of RPFs and RNA-seq for TOP3B and TDRD3 genes were reduced in TOP3B-KO1 and TDRD3-KO cells, respectively (Fig. S3a-b & Fig. S3e-f), consistent with absence of these proteins in their respective KO cells (Fig. 1b). In addition, no obvious difference was observed for TOP3B mRNA in TOP3B-KO2 cells (Fig. S3d), which is also consistent with the finding that these cells express a truncated TOP3B protein (Fig. S1c-d).
We found that the levels of RPFs and RNA-seq from both TOP3B-KO1 (3 replicates) and TOP3B-KO2 (2 replicates) cells strongly correlated with those of WT cells (R>0.9) (Fig.3a, b & Fig. S4a), indicating that inactivation of TOP3B again did not alter global translation or mRNA levels. The levels of RPFs also correlated well with those of RNA-seq (R=0.84), but poorly with those of PRO-seq (R=0.41, Fig. S4e), which are consistent with previous findings (29). The heatmaps (Fig. 5e-f) also showed strong co-clustering between RPFs and RNA-seq signals, but not with those of PRO-seq, both in the TOP3B-KO and also in TOP3B-Y336F and TDRD3-KO cells. These data suggest that translation is largely controlled by mRNA levels, but not with de novo transcription.
We identified about 1400 differentially expressed genes (DEGs, threshold >1.2-fold, p<0.05) in TOP3B-KO1 and KO2 cells by either RNA-seq or Ribo-seq (Fig. 3c, d; Table S3). A Venn diagram identified DEGs that were commonly reduced or increased in both KO1 and KO2 cells (Fig. S4f & Table S3). We used these common genes to represent the DEGs in TOP3B-KO cells in subsequent calculations to reduce the effect of clonal variation. In total, there were 884 DEGs by RNA-seq and 550 by Ribo-seq (Table S3).
Examination of the 550 DEGs identified by Ribo-seq showed that the percentage of decreased vs. increased genes were comparable (47% vs. 53%), implying that TOP3B may affect translation either positively or negatively. Because translation of an mRNA is often affected by its transcription and degradation (30), some of these DEGs could be regulated by TOP3B at those steps rather than at translation per se. To identify mRNAs that were dependent on TOP3B only for translation, we divided the decreased DEGs identified by Ribo-seq into 4 groups, based on whether these genes show concomitant decrease in RNA-seq (which reflects steady-state levels) and PRO-seq (which reflects nascent transcripts) (Fig. 3e & Table S3). Group 1 DEGs showed no concomitant decrease in RNA-seq and PRO-seq levels, which should represent genes regulated by TOP3B at translation level only. Group 2 DEGs exhibited decreased signals in RNA-seq, but no change in PRO-seq signals. This group consists of transcripts like TOP3B mRNA, whose decreased translation also reduces mRNA levels through no-go decay or nonsense-mediated decay mechanisms (31-33). The group may also contain genes of which translation reduction is due to decreased mRNA levels caused by accelerated mRNA degradation. Group 3 DEGs showed concomitant decrease in PRO-seq but no change in RNA-seq, whereas Group 4 DEGs showed decrease in both RNA-seq and PRO-seq (Fig. 3e). The last two groups likely consist of genes regulated at the transcription step.
This analysis revealed that majority of the decreased DEGs belong to group I (60%) and II (22%), suggesting that TOP3B mainly acts post-transcriptionally to regulate translation and mRNA stability. Analysis of the increased DEGs obtained similar results, as the most of the DEGs belong to post-transcriptional groups (I and II) (Fig. 3e). The number of decreased DEGs in group 1 and 2 was comparable to that of increased DEGs (212 vs. 189), supporting the notion that TOP3B can both promote and repress translation.
Several TOP3B-regulated mRNAs are important for mental disorders
Notably, the DEGs of Group 1 and 2 consist of multiple autism or schizophrenia risk genes, including CHD8, FGFR1, SGSH and FAT1 (Figure 3e-f & Table S4). BedGraph visualization confirmed the decrease or increase of the RPF signals across the gene body (Fig. 3f). Their patterns resemble that of the positive control gene (TOP3B) (Fig. S3a), suggesting that translation of these genes was disrupted in TOP3B-KO cells. The difference in RPF signals between TOP3B-KO and WT cells was significant (p<0.05) (Fig. 3g). As a control, the mRNA levels of CHD8, SGSH and FAT1 in TOP3B-KO cells exhibited no significant difference comparing to those of WT cells (Fig. 3f-g), suggesting that translation efficiency but not mRNA levels were disrupted by TOP3B inactivation. The nascent transcript levels (PRO-seq) for these genes showed no obvious difference relative to those of WT cells (Fig. 3f-g), indicating that these genes are regulated by TOP3B post-transcriptionally.
We further performed polysome profiling coupled with RT-qPCR and immunoblotting to verify changes in mRNA translation in TOP3B-KO cells. The former method measures the translation by separating the mRNAs based on the numbers of ribosomes on them through sucrose gradient (34). For CHD8 mRNA, the peak in the heavy polysome fraction in TOP3B-KO cells was decreased, whereas that in the light polysome fraction was increased (Fig. 3h), consistent with the Ribo-seq data that its translation was reduced. Immunoblotting confirmed a reduced CHD8 protein level and increased FAT1 protein level in TOP3B-KO cells (Fig. 3i-j & Fig. S5c). As controls, the level of TOP3B mRNA was decreased in the polysome fractions but increased in the monosomes in TOP3B-KO cells; and the level of GAPDH mRNA was largely unchanged (Fig. 3h & Fig. S3c). Moreover, the overall polysome profile was indistinguishable between TOP3B-KO and WT cells (Fig. S4g), supporting the findings from Ribo-seq that TOP3B inactivation does not affect global mRNA translation. Together, these data suggest that TOP3B inactivation disrupts translation of multiple mRNAs important for mental disorders.
TOP3B topoisomerase activity regulates RNA levels and translation
We analyzed TOP3B-Y336F-KI cell line to study the unresolved question of whether TOP3B depends on its topoisomerase activity for its function. The RNA-seq and Ribo-seq signals of Y336F-KI cells strongly correlated with those of WT cells (Fig. S4b; R>0.95), which resemble those of TOP3B-KO cells (Fig.3a-b), suggesting that inactivating either TOP3B topoisomerase activity or the entire protein does not affect global mRNA levels or translation. Our RNA-seq and Ribo-seq analyses identified about 1000 decreased and 2000 increased DEGs in Y336F-KI cells (Fig. 4a; Table S5). Comparison of these DEGs with those of TOP3B-KO cells revealed the percentages of commonly decreased and increased DEGs: 24% and 50%, respectively by RNA-seq; and 27% and 51%, respectively by Ribo-seq (Fig. S5a-b; Fig. S6a; Table S6-7). We then assessed whether the DEGs of TOP3B-Y336F cells showed the same direction of alteration in TOP3B-KO cells by chance using identical numbers of randomly selected genes as comparisons. We found that the observed percentages of the commonly decreased or increased DEGs are about 2.5 to 3.5 -fold greater than those of the randomly selected genes (Fig. 4b; Table S7), implying that the commonly altered DEGs do not likely occur by chance but are most probably co-regulated by TOP3B and its catalytic activity. We refer to these genes as the TOP3B-Catalytic-Activity-dependent Genes (abbreviated as TCAGs) (marked as “+” in Fig. 4c-d). Conversely, the percentages of DEGs that are altered in opposite directions between TOP3B-KO and Y336F cells are lower than those of randomly selected genes (Fig. 4b), indicating that TOP3B protein and its catalytic activity usually function in the same direction. This conclusion is further supported by evidence that the percentages of DEGs altered in the same direction are 2 to 10-fold higher than those in the oppositive direction (Fig. 4b).
Our heatmap analysis showed that a large fraction (about 50-75%) of the decreased or increased DEGs by RNA-seq or Ribo-seq in TOP3B-KO cells did not overlap with those in Y336F-KI cells (Fig. 4c-d), again suggesting that TOP3B may also regulate RNA levels and translation using mechanisms independent of its topoisomerase activity. Similarly, a large fraction (about 90%) of DEGs in Y336F-KI cells do not overlap with those of TOP3B-KO cells, hinting that the TOP3B-Y336F mutant protein may alter mRNA levels and translation in ways that are different from that caused by loss of the protein.
We also examined the representative mental disease genes that are altered in TOP3B-KO cells and found that they exhibit a similar pattern in TOP3B-Y336F cells: decreased or increased by Ribo-seq and RNA-seq analyses, and unchanged in PRO-seq analyses (Fig. 4e-f), indicating that normal translation or RNA level control of these genes depend on TOP3B topoisomerase activity. This conclusion is further supported by immunoblotting, showing similar alteration of CHD8 and FAT1 proteins between TOP3B -KO and KI cells (Fig. 3i-j & Fig. S5c).
We noticed that the levels of CHD8 and FGFR1 mRNAs by RNA-seq and Ribo-seq analyses were concomitantly reduced in Y336F-KI cells (Fig. 4e-f). The reduction of CHD8 mRNA level was further confirmed by RT-qPCR analysis (p<0.05) (Fig. S7a-b), suggesting that this mRNA may be bound and stabilized by TOP3B. We studied whether this reduction of mRNA levels is due to altered translation by pre-treating cells with a translational elongation inhibitor, cycloheximide (CHX), which is expected to increase the levels of those mRNAs subject to translation-associated mRNA decay (35). The CHD8 mRNA in TOP3B-KO and Y336F-KI, but not WT cells, was significantly increased by CHX treatment (Fig. S7c-e), indicating that the reduced level of mRNA is due to reduced translation caused by inactivation of TOP3B catalytic activity. As a positive control, TOP3B mRNA level was also increased by CHX treatment in TOP3B-KO1 but not WT cells (Fig. S7c). This is expected because the TOP3B frameshift mutations in KO1 cells create premature stop codons in the earlier exons, making this mRNA a substrate of NMD that can be stabilized by inhibiting translation elongation (35). In summary, these data suggest a hierarchical action: TOP3B topoisomerase activity regulates translation of an mRNA, and this process stabilizes the mRNA.
TOP3B co-regulates mRNA levels and translation with TDRD3 and FMRP
We studied whether TOP3B and its two partners, TDRD3 and FMRP, co-regulate a common set of genes. We analyzed TDRD3-KO and FMR1-KO cells using the same methods described above. The levels of RNA-seq and Ribo-seq of the two KO cells strongly correlated with those of WT cells (R>0.95; Fig. S4c-d). These features resemble those of TOP3B-KO cells, suggesting that inactivation of TOP3B-TDRD3 and FMRP affects neither global mRNA levels nor translation. Volcano plots identified approximately 2500 DEGs in TDRD3-KO cells by each assay; and about 800~1300 in FMR1-KO cells (Figure 5a-d). The number of DEGs in TDRD3-KO cells is about twice those of TOP3B-KO and FMR1-KO cells (Fig. 3c-d), consistent with the earlier data that TDRD3 can function independently of the TOP3B-TDRD3 complex (36).
A Venn diagram showed that the number of commonly decreased or increased genes between TOP3B-KO and TDRD3-KO is larger than that between TOP3B-KO and FMRP-KO cells (Fig. S6b-c). Heatmaps revealed that there are more commonly downregulated (blue) or upregulated (37) DEGs between TOP3B-KO and TDRD3-KO than those between TOP3B-KO and FMRP-KO in both RNA-seq and Ribo-seq assays (Fig. 5e-f; compare column 4 and 7). For example, the percentage of commonly decreased DEGs between TOP3B-KO and TDRD3-KO cells is about 2-fold higher than those between TOP3B-KO and FMR1-KO in RNA-seq (37% vs. 18%) and Ribo-seq (56% vs. 27%) (Fig. 5e-f). The data suggest that TOP3B coregulates more genes with TDRD3 than with FMRP. The findings correlate with the interaction data that TOP3B forms a stoichiometric complex with TDRD3, and only a minor fraction of this complex interacts with FMRP (2, 3, 38).
We then assessed whether the observed percentages of DEGs of TOP3B-KO cells showing the same alteration in TDRD3-KO or FMR1-KO cells could happen by chance, using randomly selected genes as control. The results showed that the observed percentages are 3 to 5-fold or 2 to 3-fold higher respectively, than those of control genes in TDRD3-KO or FMR1-KO cells, indicating that these concomitantly altered DEGs are less likely to occur by chance but are more probably genes co-regulated by TOP3B and its partners (Fig. 5g-h). In contrast, the percentages of DEGs showing the opposite alterations between TOP3B-KO and TDRD3-KO were lower than those of randomly selected genes, indicating that TOP3B and TDRD3 largely acts in the same direction in RNA level control and translation (Fig. 5g-h). This conclusion is further supported by that the percentages of DEGs showing the same direction of alteration are about 6-10-fold greater than those of DEGs showing opposite direction (Fig. 5g-h). Similarly, the percentages of DEGs showing the oppositive changes between TOP3B-KO and FMR1-KO are roughly equal to those of randomly selection genes, hinting that these changes could occur by chance, and these genes may not be co-regulated by TOP3B and FMRP (Fig. 5g-h). The percentages of DEGs altered in the same direction between TOP3B-KO and TDRD3-KO cells were about 3-fold higher than those between TOP3B-KO and FMR1-KO, suggesting that TOP3B and TDRD3 coregulate more genes than TOP3B and FMRP.
We also analyzed DEGs of FMR1-KO cells and found a stronger overlap with those of TDRD3-KO than TOP3B-KO or Y336F cells, in both RNA-seq (downregulated: 39% vs. 32% or 35%, respectively) and Ribo-seq (downregulated: 38% vs. 20% or 21%, respectively) (Fig. S8a-b). The percentages of concomitantly decreased or increased DEGs between FMR1-KO and TDRD3-KO or TOP3B-KO were about 2 to 5-fold higher than those of randomly selected genes, whereas those of DEGs that are altered in opposite directions were either similar or lower than those of randomly selected genes (Fig. S8c). The data support the notion that TOP3B-TDRD3 and FMRP co-regulate mRNA levels and translation.
We examined the four mental disease genes with altered expression in TOP3B-KO and Y336F mutant cells and found that they all display the same direction of alteration in TDRD3-KO cells, whereas only two of the four exhibit the same trend in FMR1-KO (FGFR1 and FAT1) (Fig. 5i-j). Together, these data suggest that TOP3B co-regulates more mRNAs in conjunction with TDRD3 than with FMRP at post-transcriptional steps.
TOP3B-TDRD3 regulates mRNAs in both topoisomerase-dependent and independent mechanisms
We hypothesized that the co-regulated mRNAs in both TOP3B-KO and TDRD3-KO cells (Fig. 6a-b & Table S8) are controlled by the entire TOP3B-TDRD3 complex and investigated how these mRNAs are regulated by the topoisomerase activity and FMRP. Comparing to WT cells, the percentage of decreased mRNAs is roughly equal to the increased mRNAs in RNA-seq (49% vs. 51%) and about 50% fewer (40% vs. 60%) in Ribo-seq (Fig. 6a-b), suggesting that the TOP3B-TDRD3 complex can affect gene expression either positively or negatively. There is strong overlap between RNA-seq and Ribo-seq data in both TOP3B-KO and TDRD3-KO cells: 64%-88% for the decreased DEGs, and 77%-93% for increased DEGs, suggesting a coordinated regulation by TOP3B-TDRD3 (Fig. 6a-b).
To determine whether the complex acts transcriptionally or post-transcriptionally, we analyzed PRO-seq levels for these DEGs. For the decreased DEGs from both RNA-seq and Ribo-seq assays, less than 38% exhibit concomitant decrease in PRO-seq levels (Fig. 6a-b). This contrasts with the increased DEGs, in which more than 51% exhibited concomitant increase in PRO-seq levels. The data resembles findings above (Fig. 5e-f) and suggests that the TOP3B-TDRD3 complex suppresses gene expression at the transcriptional level but enhances gene expression mainly post-transcriptionally.
We analyzed the decreased DEGs of the 2 KO cell lines and found that more than 40% of them were concomitantly decreased in Y336F-KI cells, whereas about 30% were decreased in FMR1-KO cells (Fig. 6a-b). The data thus suggest that some but not all mRNAs regulated by the TOP3B-TDRD3 complex are under control of the topoisomerase activity of TOP3B and FMRP.
The mRNAs altered in both TOP3B-KO and TDRD3-KO cells include 44 mental disorder-related genes (Fig. 6c). Half of them (22) were also reduced in Y336F cells. In addition to the four representative transcripts (CHD8, FGFR1, SGSH and FAT1) that were altered in Y336F cells (Fig. 4e), three more transcripts (PRR12, SMC3 and BCORL1) were unchanged in Y336F cells (Fig. 6d-e). These findings further reinforce the notion that the TOP3B-TDRD3 complex can regulate the expression of genes important for mental disorders in topoisomerase activity dependent and independent manners.
TOP3B preferentially binds long mRNAs enriched in mental disorder genes
We performed eCLIP-seq to identify TOP3B-bound mRNAs and determine whether they overlap with those regulated by TOP3B-TDRD3 described above (22). We identified 1106 TOP3B-bound mRNAs based on two independent experiments in HCT116 cells (Table S9). About 1/3 of these mRNAs matched those identified from HeLa cells by TOP3B HITS-CLIP(3) (Fig. S9a). Those with higher eCLIP-seq signals tend to have higher HITS-CLIP signals (Fig. 7a), suggesting that TOP3B may recognize specific features of these mRNAs. We found several features that are common in TOP3B-bound mRNAs identified in the current and previous studies, which are also shared by FMRP-bound mRNAs. First, the largest fraction of TOP3B CLIP-reads (~50%) are localized in coding regions of mRNAs (Fig. 7b) (3), consistent with findings that TOP3B-TDRD3 associates with polyribosomes and regulates translation. Second, the average lengths of these mRNAs are significantly longer than that of randomly selected unbound mRNAs (Fig. 7c) (3, 39). This also resembles findings that TOP1 and TOP2-regulated genes tend to have longer average lengths and are enriched in genes important for autism and neurological disorders (40). Third, a fraction of TOP3B-bound mRNAs (17%) from both current and previous studies overlapped with those bound by FMRP (Fig. S9b) (41), including many autism (https://gene.sfari.org/), and schizophrenia-related (http://bioinfo.mc.vanderbilt.edu/SZGR/) mRNAs (109 and 59, respectively) (Fig. 7d-e; Table S9), supporting the proposal that TOP3B may work with FMRP to regulate translation of mRNAs important for mental disorders (3). To verify that TOP3B can indeed bind these mRNAs, we conducted RNA immunoprecipitation (RIP) analysis coupled with RT-qPCR and found that TOP3B antibody immunoprecipitated CHD8 and FAT1mRNAs, but not GAPDH or ACTB mRNAs (Fig. S9g). The data indicate that TOP3B can directly bind and regulate mRNAs important for autism or schizophrenia.
TOP3B binding but not catalytic activity stabilizes target mRNAs
We next examined the effects of TOP3B inactivation on its bound mRNAs by RNA-seq or Ribo-seq analyses. In both cases, the levels of RNA and RPFs in TOP3B-KO cells strongly correlate with those of WT cells (Fig. S9c); and a minor fraction of TOP3B-bound mRNAs (12% or less; 135 out of 1106) displayed altered RNA-seq levels in TOP3B-KO cells treated with CHX to inhibit translation (Fig. 7f). Interestingly, among those with altered mRNA levels, 96% (130) were decreased, whereas only 4% (5) were increased (Fig. 7f & 7h). Only 8% of the decreased DEGs showed this trend by PRO-seq analysis (Fig. 7h), indicating that this decrease in RNA-seq signals is not due to reduced transcription. The cells without CHX treatment showed only a mild trend (about 60% decreased) (Fig. S9e). The data suggest that TOP3B binding preferentially stabilizes its target mRNAs when translation elongation is inhibited. In support of this notion, when cells were treated with CHX, TOP3B-bound mRNAs exhibited overall reduced RNA-seq signals in TOP3B-KO cells than WT cells (Fig. S9d).
We then analyzed these TOP3B-bound mRNAs in Y336F cells but did not observe a similar overall decrease (Fig. S9d & 9f), suggesting that stabilization of these mRNAs is independent of the topoisomerase activity. As examples, two representative mRNAs, AGRN and KIAA0100, were bound by TOP3B; and their RNA-seq, but not PRO-seq, signals were significantly reduced in both TOP3B-KO and TDRD3-KO, but not in Y336F-KI cells (Fig. 7j-k), indicating that they are stabilized by TOP3B-TDRD3 binding but not topoisomerase activity.
Our analysis of TOP3B-bound mRNAs in Ribo-seq found that the numbers of mRNAs showing decreased or increased signals are about equal (37 vs. 34) (Fig. 7g and 7i), suggesting that TOP3B binding may either enhance or suppress translation of its target mRNAs. In addition, we found that 30 of the TOP3B-bound mRNAs display altered Ribo-seq signals in Y336F-KI cells. Among them, the number of mRNAs showing decreased abundance (11) is about 1.7-fold lower than those with increased levels (19) (Table S10), suggesting that TOP3B topoisomerase activity may either enhance or suppress translation of its bound mRNAs.