Dbp7 interacts with RNA exosome component Dis3 to mediate CENP-A loading to centromeres

Abstract

Background

Centromeres are crucial for proper chromosome segregation during cell division. Centromeres in most eukaryotes are epigenetically defined by the histone H3 variant, CENP-A. Centromeric regions are typically transcribed into non-coding RNAs that contribute to centromere functions. However, the precise role of centromeric RNAs in CENP-A loading, in particular the formation of R-loops, remains poorly understood and the mechanisms that safeguard centromeres from R-loop-associated defects are still largely unclear.

Results

Using a visual genetic screen in fission yeast, we find that a conserved ATP-dependent RNA helicase Dbp7, an RNA binding protein that plays a role in preventing accumulation of R-loops at centromeres, is a critical regulator of CENP-A loading to centromeres. Deletion of dbp7⁺ results in CENP-A mislocalization, defects in chromosome segregation, and impaired centromere silencing. Domain-deletion analysis demonstrates that both DEAD-box helicase domain and the C-terminal extension motif contribute to its centromere function. We further show that its catalytic activity is crucial for its role at centromeres. Mechanistically, Dbp7 interacts with the key exosome subunit Dis3, which promotes CENP-A centromere loading by interacting with the CENP-A chaperone Sim3.

Conclusions

Together, this study uncovers a previously unrecognized CENP-A loading mechanism, by which Dbp7 and Dis3 act together to bind to centromeric transcripts and in turn mediate recruitment of CENP-A via the CENP-A chaperone Sim3, providing mechanistic insight into R-loop resolution at centromeres.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13059-026-04133-8.

Background

The centromere is a specialized chromatin domain responsible for kinetochore assembly, which is essential for faithful chromosome segregation during mitosis and meiosis. Most eukaryotes contain large and complex “regional” centromeres, which span a large region on the chromosome, allowing for multiple microtubule attachment sites per centromere [1, 2]. Regional centromeres are epigenetically specified by the conserved centromere-specific histone 3 (H3) variant, Centromere protein-A (CENP-A). CENP-A nucleosomes, interspersed with the canonical histone H3 within centromeres, directly interacts with the kinetochore subcomplex, CCAN (constitutive centromere-associated network) and act as the primary recruitment site for kinetochore assembly [1–4]. Mislocalization of CENP-A in multiple organisms results in ectopic assembly of CENP-A chromatin and severe chromosome segregation defects [5–9]. Loading of CENP-A to centromeres is mediated by the conserved CENP-A-specific chaperone, HJURP/Scm3 [10–15]. Recruitment of HJURP requires the conserved Mis16-Mis18 complex [16, 17]. CCAN subunits, such as CENP-C and CENP-I, are also important for CENP-A assembly at centromeres [18–22]. How CENP-A is precisely targeted to centromeres remains unclear.

Centromeres in most eukaryotes are transcribed into non-coding RNAs (ncRNAs), albeit typically at low level [23, 24]. Recent studies suggest that both non-coding centromeric RNAs and the act of centromere transcription itself are important for centromere activity and kinetochore assembly [23–31]. The exact role of centromeric non-coding RNA remains poorly understood. In addition, centromere transcription, like coding gene transcription, can induce formation of R-loops, which are a three-stranded nucleic acid structure consisting of a stretch of RNA:DNA hybrids and an unpaired, single stranded DNA loop [32]. R-loops can function as regulators of gene expression, DNA damage repair and DNA replication. But excessive and persistent R-loops can lead to DNA damage and genome instability [33]. A recent study showed that Hpr1, a component of the THO/TREX complex, prevents the accumulation of R-loops at centromeres for chromosomal stability in Saccharomyces cerevisiae [34]. The mechanism used for safeguarding centromeres against R-loops is still little known.

The fission yeast (Schizosaccharomyces pombe) has long been a model organism for centromere studies. Unlike S. cerevisiae that contains a point centromere defined by a short centromere sequence, S. pombe contains large regional centromeres that are epigenetically defined by the CENP-A homolog, Cnp1 [21, 35]. Cnp1-containing regions consist of the non-repetitive centromere core (cnt) and flanking innermost repeats (imr) in centromeres. Scm3^HJURP and the inner kinetochore CENP-I homolog Mis6 in S. pombe promote Cnp1 loading to centromeres [13, 14, 21]. In addition, a conserved NASP family protein Sim3 is also required for Cnp1 loading by interacting with Cnp1 [36]. Similar to higher eukaryotes, centromeres in S. pombe are transcribed but weakly silenced [37, 38]. Proper transcription in fission yeast centromeres facilitates the deposition of CENP-A [39, 40]. The centromere transcripts are subject to rapid turnover by the exosome subunit Dis3 [38]. It has been shown that the Mis6 subcomplex inhibits the indiscriminate transcription of non-coding RNAs at the core centromere [38, 41]. Like many eukaryotes, centromeres in S. pombe are flanked by more strongly silenced pericentromeric heterochromatin [42, 43]. The pericentromeric heterochromatic regions are characterized by H3K9 methylation (H3K9me) [44, 45]. Pericentromeric heterochromatin transcripts can be processed by both RNA interference (RNAi) and the exosome pathways, which in turn promote H3K9me and heterochromatin silencing [46–48]. Nevertheless, H3K9me and RNAi do not appear to be involved in CENP-A-containing centromeric chromatin [49]. How non-coding centromeric RNAs are regulated and its impact on CENP-A loading remains elusive.

Here we show that through a visual genetic screen, we identified a conserved but poorly characterized RNA binding protein Dbp7 that plays a critical role in CENP-A loading to centromeres. Deletion of dbp7⁺ results in CENP-A mislocalization and chromosome segregation defects. Centromere silencing is also lost in the dbp7∆ mutant. Dbp7 is primarily localized in the nucleus, and interacts with centromere transcripts. Importantly, Dbp7 is important for preventing accumulation of R-loops at centromeres. We further showed that its catalytic activity is crucial for its role in centromere function. Rnh1 is an RNase H enzyme important for R-loop elimination [50]. We also found here that both deletion and overexpression of Rnh1 leads to CENP-A mislocalization and loss of centromere silencing, consistent with the idea that maintaining proper R-loop levels in centromeres is essential for centromere function. Nevertheless, centromere defects in rnh1Δ are significantly weaker than dbp7∆, suggesting that Dbp7 plays an additional role in CENP-A incorporation. Indeed, we found that Dbp7 interacts with the exosome subunit Dis3 that is required for CENP-A loading by interacting with the CENP-A chaperone Sim3. Together, our data suggest that Dbp7 and Dis3 bind to centromeric transcripts, which in turn mediate CENP-A loading to centromeres via recruiting Sim3. This study uncovers a mechanism for how RNA quality control factors promote CENP-A positioning at centromeres and provides mechanistic insight into how centromeres are safeguarded against harmful R-loops.

Results

Deletion of dbp7⁺ results in abnormal CENP-A distribution and chromosome mis-segregation

We previously initiated a genomic-wide genetic screen to identify factors important for CENP-A loading to centromeres using Cnp1-GFP as a visual marker. In S. pombe interphase cells, three centromeres are clustered underneath the nuclear envelope; thus, a single Cnp1-GFP focus is detected in wild-type (WT) cells expressing Cnp1-GFP at its native level [43, 51] (Fig. 1A). We crossed WT strain with Cnp1-GFP with individual mutants from the S. pombe genomic-wide deletion library and visually examined the resulting strains. Through this screen, we have previously identified Ccp1 and the MBF complex important for proper Cnp1 positioning [52–54]. In addition, we found that deletion of the uncharacterized dbp7⁺ gene results in abnormal CENP-A distribution. Overall, the GFP signal in the dbp7Δ cells is significantly dimmer than WT cells in the whole population, indicating that the level of Cnp1-GFP at centromeres is reduced (Fig. 1A and B). In addition to the dimmer GFP signal, dbp7Δ cells often display multiple foci of Cnp1-GFP (Fig. 1C). We confirmed that the mutant cells carrying multiple Cnp1-GFP foci are interphase cell using the spindle pole body (SPB) marker, Sad1-mCherry, which displays a single focus during interphase (Fig. 1D) and found that more than 10% dbp7Δ interphase cells contain multiple Cnp1-GFP foci (1E). These data indicated that Cnp1 is also mislocalized to the ectopic regions.

Fig. 1.

The dbp7∆ mutant exhibits abnormal distribution of CENP-A. A Cnp1-GFP signal in the dbp7∆ mutant is weaker than WT. Bottom right panel: higher exposure time for dbp7∆ was also shown. Equal exposure, 0.2 s. Higher exposure, 0.8 s. Scale bars, 2 μm. B Quantification of Cnp1-GFP signal intensity in WT and dbp7∆ cells. n = 40. Error bars represent mean and SD. ***, P < 0.001. C The dbp7∆ mutant cells contain multiple Cnp1-GFP foci. Scale bars, 2 μm. D Excessive Cnp1-GFP foci were visualized in the dbp7∆ mutant interphase cells carrying the SPB marker, Sad1-mCherry. WT cells with Sad1-mCherry were used as a control. Scale bars, 2 μm. E The frequency distribution of the number of Cnp1-GFP foci per cell (> 1) in the both WT and dbp7∆ mutant interphase cells. n = 40. F Genome browser views showing ChIP-seq analysis of Cnp1-GFP distribution in dbp7∆/WT across three chromosomes. The dbp7∆/WT is the log2 ratio of Cnp1-GFP signal between the dbp7∆ mutant and wild type plotted across chromosomes. ChIP was performed using an antibody against GFP

To further verify the mislocalization of Cnp1 in dbp7Δ, we analyzed dbp7Δ cells carrying Cnp1-GFP by ChIP followed by high-throughput sequencing (ChIP-seq) with an antibody specific for GFP. WT cells carrying Cnp1-GFP were used as a control. Our ChIP-seq data showed that Cnp1-GFP is significantly enriched in euchromatic regions but decreased at centromeres in the dbp7Δ mutant relative to WT (Fig. 1F). Together, our results indicate that Dbp7 is important for proper Cnp1 loading at centromeres.

Dbp7 is predicted to be an RNA binding protein conserved across eukaryotes, but it has not been well-studied. Recent studies revealed that Dbp7 in S. cerevisiae belongs to the DEAD-box family of ATP-dependent RNA helicases [55, 56]. Its S. cerevisiae and human homologs have been shown to participate in the transcription of rRNA genes and the assembly of ribosome subunits [57, 58]. DDX31, the Dbp7 human homolog, also regulates transcriptional activation of mutant p53 (mutp53) in bladder cancer cells, and has been implicated in carcinogenesis [59].

To determine whether loss of dbp7⁺ leads to chromosome instability (CIN), we examined the growth phenotype of dbp7∆ cells exposed to microtubule inhibitor thiabendazole (TBZ). TBZ selectively binds to ß-tubulin and inhibits polymerization and its assembly into microtubules. The mutants with TBZ sensitivity demonstrate defects in chromosome segregation. Our results showed that dbp7∆ has strong sensitivity to TBZ compared to WT cells, indicating chromosome mis-segregation in the mutant (Fig. 2A). To directly examine the mitotic chromosome segregation, we visualized the spindle by using tubulin-GFP, and counterstained DNA with DAPI. We frequently observed the unequal chromosome segregation or lagging chromosome in the mutant during mitosis (Fig. 2B and C). The chromosome segregation defects in the dbp7Δ mutant were further confirmed by the minichromosome loss analysis [60] (Additional file 1: Fig. S1). We also found that dbp7∆ cells exhibit irregular, often elongated cell shapes (Fig. 2D and E). The abnormal shape of the mutant could result from chromosome mis-segregation during cell division.

Fig. 2.

The dbp7∆ mutant is defective in chromosome segregation. A dbp7∆ mutant cells are sensitive to TBZ. Serially diluted cells were plated on rich YES media containing 5 μg/ml, 10 μg/ml, or 15 μg/ml TBZ. B dbp7∆ cells exhibit uneven or lagging chromosomes during mitosis. Segregation of chromosomes in WT and dbp7Δ mitotic cells were visualized by DAPI staining (red) and tubulin-GFP (green). Scale bars, 2 μm. C Quantification of the proportion of cells displaying a segregation defect in indicated strains. n = 40. D dbp7∆ mutant cells display defect cell shapes, as shown by the bright-field microscopy image. E Quantification of cell length in the indicated strains. Data are shown as Mean ± SD (n = 40). ***, P < 0.001

Dbp7 is required for centromere silencing

To investigate the role of dbp7⁺ in centromere silencing, we examined the silencing level of the centromere core region in the dbp7∆ mutant. We took advantage of a strain carrying a ura4⁺ gene integrated at the centromeric core cnt1 domain (cnt1::ura4⁺) [37]. The ura4⁺ gene is required for biogenesis of uracil, which is essential for cell growth. We crossed dbp7∆ into the strain carrying cnt1::ura4⁺, and the resulting strain and WT as a control were analyzed on the media containing FOA (5-fluoroortic acid), a counter-selective drug toxic to cells expressing ura4⁺. In WT strain, the ura4⁺ gene is transcriptionally silenced, which results in fast growth on the FOA media. However, the cells without dbp7⁺ showed cannot grow on the FOA plate, indicating that centromeric silencing is lost (Fig. 3A). Consistent with this, our ChIP-qPCR assays showed that the histone acetylation level is increased in the centromere core region in dbp7∆ (Fig. 3B).

Fig. 3.

Transcriptional silencing at centromeres is lost in dbp7∆. A tenfold serial dilutions of indicated cells harboring ura4⁺ at the centromere core domain (cnt) region were plated on the counterselective FOA medium and incubated at 30 °C for 4 days. ctrl, minimal media without FOA. B ChIP-qPCR assays showed that H3K9ac in the cnt region in dbp7∆ mutant cells were significantly increased. The data were normalized to the input. The Y-axis represents the fold change and the level of WT was set to 1. Data are shown as Mean ± SD (n = 3). *, P < 0.05. C Growth assays of cells carrying dbp7∆ combined with mis6-302 compared with single mutants grown on YES media at 25, 32, or 36 °C

Mis6 is a key player in CENP-A loading, and its mis-regulation results in defects in CENP-A localization and also loss of silencing in centromeres [21, 38, 41]. To determine whether Dbp7 acts in the same pathway as Mis6, we utilized the mis6-302 temperature-sensitive mutant, which grows normally at room temperature but not 36 °C. We created the double dbp7∆ mis6-302 mutant. Our growth assays at different temperatures showed that there is no detectable synthetic defect in the double mutant (Fig. 3C), suggesting that Dbp7 and Mis6 act in the same pathway.

Dbp7 associates with centromere transcripts

To determine the distribution pattern of Dbp7 in cells, we replaced the endogenous Dbp7 with TAP-tagged Dbp7 at its C terminus by homologous recombination, and conducted immunostaining of cells carrying Dbp7-TAP using an antibody against the TAP tag and DAPI. To make sure that the TAP tag does not interfere with Dbp7 function, we crossed cells expressing Dbp7-TAP into the cnt1::ura4⁺ background and performed the growth assays on the FOA media. We found that Dbp7-TAP cells grow at the same rate as WT with cnt1::ura4⁺, indicating that Dbp7-TAP is functional (Additional file 1: Fig. S2). Our immunostaining assays showed that Dbp7-TAP is primarily localized in the nucleus, and often accumulates at the darker, less intensely DAPI-stained area within the nucleus (Fig. 4A), suggesting that Dbp7 is enriched in the nucleolus. We further confirmed that the enrichment of Dbp7 in the nucleolus by immunostaining analysis of cells expressing S-tagged Dbp7 and the nucleolus marker Nuc1-GFP (Additional file 1: Fig. S3). In addition, Dbp7-TAP also formed distinct fluorescence foci in the nucleoplasm near the nuclear periphery (Fig. 4A), consistent with the idea that Dbp7 associates with centromeres.

Fig. 4.

Dbp7 is enriched in the nucleus and associates with centromere transcripts. A Cells carrying Dbp7-TAP were stained with anti-TAP (green) and DAPI (blue). White arrows denote Dbp7-TAP foci in the nucleoplasm. Scale bar, 2 μm. B RIP assay shows that Dbp7 associates with centromere transcripts. (Left panel) RIP was performed from extracts of mis6-302 cells expressing Dbp7-TAP using an anti-TAP antibody. Untagged mis6-302 cells were used as a control. Cells were incubated at 36 °C for 8 h before the RIP assays. (Right panel) RIP analysis of mis6-302 cells expressing the tubulin protein Tub1-GFP using an anti-GFP antibody as a negative control. Immunoprecipitated RNAs were quantified by RT-qPCR using primers specific for the cnt region. The data were normalized to the input. The Y-axis represents the fold change and the RNA level immunoprecipitated by control RIP was set to 1. Experiments were performed in triplicate. Data are shown as Mean ± SD. **, P < 0.01. ns = statistically not significant

Dbp7 is predicted to be an RNA-binding protein. We then tested whether Dbp7 physically binds to centromeric transcripts. Since centromeric RNA level is low in WT cells due to exosome processing, we took advantage of the mis6-302 mutant, which contains high-levels of centromeric RNA at restrictive temperature [38]. We performed the RNA Immunoprecipitation (RIP) assays using mis6-302 cells containing Dbp7-TAP with an anti-TAP antibody. The mis6-302 mutant without TAP-tag was used as a control. These strains were incubated at restrictive 36 °C for 8 h before the RIP assays. Our RIP results showed that Dbp7-TAP was copurified with centromeric transcripts (Fig. 4B). As a control, we also performed RIP analysis of mis6-302 cells carrying the tubulin protein Tub1-GFP using an GFP antibody. We found that Tub1 has no significant enrichment with centromeric RNAs (Fig. 4B). Our data indicate Dbp7 interacts with centromeric RNAs and further suggest that Dbp7 regulates centromere function through interacting with centromeric transcripts.

Dbp7 prevents accumulation of R-loops at centromeres

Centromere transcription can lead to formation of R-loops. Excessive R-loops in centromeres give rise to DNA damage and genome instability [32]. Rnh1 is an RNase H enzyme in fission yeast and plays an important role in R-loop elimination [50]. To determine the role of R-loops in centromere function in S. pombe, we first examined the centromere silencing in the rnh1Δ cells carrying ura4⁺ at the cnt1 domain on the FOA media. We found that rnh1Δ grew significantly slower than WT, indicating that centromere silencing is lost in the mutant (Fig. 5A). We also examined Cnp1-GFP distribution in rnh1Δ, and found that Cnp1-GFP signal is decreased in centromeres (Fig. 5B and C). However, the Cnp1-GFP distribution defect in rnh1Δ is significantly weaker than dbp7∆ (Fig. 1A and B), indicating that Dbp7 plays a more important role in CENP-A loading. Consistent with this, dbp7Δ showed a stronger centromere silencing defect than rnh1Δ (Fig. 5D). To determine the effect of overexpression of Rnh1 on CENP-A distribution, we overexpressed Rnh1 under the control of thiamine-regulated nmt1 (no message in thiamine) promoter, and found that overexpression of Rnh1 also results in substantial reduction of Cnp1 in centromeres (Fig. 5E and F). This was further supported by our data showing that overexpression of Rnh1 also results in loss of centromere silencing (Fig. 5G). These data agree with the notion that maintaining proper R-loop levels in centromeres is essential for centromere function.

Fig. 5.

Accumulation of R-loops at centromeres in dbp7∆. A Growth assays for the rnh1∆ mutant carrying ura4⁺ in the centromeric cnt region. Serial dilutions of cells were plated on the FOA medium. ctrl, minimal media without FOA. B Cnp1-GFP distribution in rnh1∆. Scale bars, 2 μm. C Quantification of Cnp1-GFP signal intensity in WT and rnh1∆ cells n = 40. Error bars represent mean and SD. ***, P < 0.001. D Growth assays for indicated strains carrying ura4⁺ in the centromeric cnt region. Serial dilutions of cells were plated on the FOA medium. ctrl, minimal media without FOA. E Cnp1-GFP distribution in cells overexpressing Rnh1. ctrl, control. OE, overexpression. Scale bars, 2 μm. F Quantification of Cnp1-GFP signal intensity in cells overexpressing Rnh1. n = 40. Error bars represent mean and SD. ***, P < 0.001. G Overexpression of Rnh1 results in loss of centromere silencing. Cells carrying rnh1⁺ under the nmt1 promoter and ura4⁺ in the centromeric cnt region were plated on the media FOA medium without thiamine (-thiamine). ctrl, minimal media without FOA and thiamine. H DRIP analysis of indicated strains using the S9.6 antibody followed by qPCR with primers specific for the cnt region. I DRIP analysis of indicated strains with treatment with RNase H. The data were normalized to the input. The Y-axis represents the fold change, and the R-loop level of WT was set to 1. Experiments were performed in triplicate. Data are shown as Mean ± SD (n = 3). ***, P < 0.001. ns, statistically not significant

Several DEAD-box helicases have been reported to prevent R-loop accumulation [61]. To investigate whether Dbp7 has a role in R-loop formation in centromeres, we analyzed both WT and the dbp7∆ mutant using DNA-RNA Immunoprecipitation (DRIP) experiments followed by qPCR with primers specific for the centromere core cnt to determine the level of R-loops in centromeres in these strains. Our DRIP-qPCR assays revealed a significant accumulation of R-loops at cnt in the dbp7∆ strain compared with WT (Fig. 5H). As a control, we also conducted the DRIP-qPCR experiments after we treated DNA with ribonuclease RNase H, which degrades RNA in DNA-RNA hybrids and results in R-loop elimination. As expected, we cannot detect R-loops at the cnt region in both WT and dbp7∆ strains after treatment of RNase H (Fig. 5I). We also created the dbp7∆ rnh1∆ double mutant carrying ura4⁺ at the cnt1 domain. Our growth assays on the FOA media showed that the double mutant exhibits a similar centromere silencing defect as the single dbp7∆ mutant (Fig. 5D), suggesting that the two genes act in the same pathway. Together, these results demonstrate that Dbp7 can prevent the accumulation of R-loops at centromeres.

Middle- and C-terminal domains of Dbp7 contribute to its role in centromere function

Dbp7 has a modular organization consisting of an N-terminal disorder region, a middle DEAD/DEAH box helicase domain, and a C-terminal extension motif. To further understand the role of Dbp7, we generated a series of domain-deletion mutants of dbp7⁺, in which each domain was deleted systemically by homologous recombination (Fig. 6A). To determine the role of each domain in centromere function, we crossed each domain-deleted mutant into the cnt1::ura4⁺ background and conducted growth assays on the FOA media. We found that whereas the N-terminal deletion mutant, dbp7-∆N, displayed little silencing defect, the silencing in the middle domain- and C-terminal deletion mutants, dbp7-∆M and dbp7-∆C, respectively, was severely lost (Fig. 6B). These data indicate that the middle- and C-terminal regions contribute to its function in centromere silencing.

Fig. 6.

Domain-deletion analysis showed the middle- and C-terminal domains of Dbp7 important for its centromere function. A A schematic diagram of Dbp7 domain-deletion mutants. B Growth assays for Dbp7 domain-deletion mutants carrying ura4⁺ in the centromeric cnt region. Serial dilutions of cells were plated on the FOA medium. C DRIP analysis of indicated strains using the S9.6 antibody followed by qPCR with primers specific for the cnt region. D DRIP analysis of indicated strains with treatment with RNase H. The data were normalized to the input. The Y-axis represents the fold change, and the R-loop level of WT was set to 1. Experiments were performed in triplicate. Data are shown as Mean ± SD (n = 3). *, P < 0.05. ns, statistically not significant

To determine the role of different domains in R-loop formation, we analyzed the domain-deletion mutants of Dbp7 by DRIP. Our results showed that deletion of middle- and C-terminal domains of Dbp7 resulted in significant enrichment of R-loops in the centromere region, whereas deletion of N-terminus had little impact in the region (Fig. 6C and D), consistent with the silencing assay results above. These data indicate that the middle- and C-terminal regions of Dbp7 are important for its function in centromere assembly.

The catalytic activity of Dbp7 is crucial for its role in centromeres

Dbp7 is a conserved ATP-dependent RNA helicase, possessing ATPase activity [55, 56]. To determine if the ATPase activity of Dbp7 contributes to its centromere function, we created a catalytic-dead point mutant, dbp7-E302Q, in which glutamate (E) in the conserved catalytic DExD motif was substituted with glutamine (Q) (Fig. 7A). Same mutation in budding yeast Dbp7 has been shown to completely abolish its ATPase activity [55]. To explore the role of helicase activity of Dbp7 in centromere silencing, we crossed the dbp7-E302Q mutant to the WT strain carrying a ura4⁺ at the centromeric core cnt1 domain. The resulting strain were subsequently analyzed on the media containing FOA. WT and dbp7Δ carrying the same reporter were used as positive and negative control, respectively. We found that, similar to dbp7Δ, dbp7-E302Q cells cannot grow on the FOA plate (Fig. 7B), indicating that the ATPase activity is important for centromere silencing.

Fig. 7.

The catalytic activity of Dbp7 is important for its centromere function. A A schematic diagram of the Dbp7-E302Q point mutant. *, mutated amino acid. B Growth assays for the dbp7-E302Q mutant carrying ura4⁺ in the centromeric cnt region. Serial dilutions of cells were plated on the FOA medium. ctrl, minimal media without FOA. C Cnp1-GFP distribution in dbp7-E302Q. Bottom right panel: higher exposure time for dbp7-E302Q was also shown. Equal exposure, 0.2 s. Higher exposure, 0.8 s. Scale bars, 2 μm. D Quantification of Cnp1-GFP signal intensity in WT and dbp7-E302Q cells. n = 40. Error bars represent mean and SD. ***, P < 0.001. E TBZ sensitivity analysis of dbp7-E302Q. Serially diluted cells were plated on rich YES media containing 5 μg/ml, 10 μg/ml, or 15 μg/ml TBZ. F DRIP analysis of indicated strains using the S9.6 antibody followed by qPCR with primers specific for the cnt region. The data were normalized to the input. The Y-axis represents the fold change, and the R-loop level of WT was set to 1. Experiments were performed in triplicate. Data are shown as Mean ± SD (n = 3). **, P < 0.01

We next investigated the Cnp1-GFP distribution in dbp7-E302Q. We found that GFP signal at centromeres in the mutant cells is considerably weaker than WT cells (Fig. 7C and D). In addition, dbp7-E302Q cells often display multiple foci of Cnp1-GFP (Fig. 7C). These data indicate that lack of the catalytic activity of Dbp7 leads to significant Cnp1 mislocalization. Consistent with this, dbp7-E302Q cells show strong chromosome segregation defects, as evident by its high sensitivity to TBZ (Fig. 7E). We next investigated R-loop level in dbp7-E302Q by DRIP, and found that R-loops are significantly enriched in centromeres of the mutant (Fig. 7F), indicating that the ATPase activity of Dbp7 is important for R-loop resolution in centromeres.

Dbp7 interacts with Dis3 to mediate CENP-A loading

A previous proteomic study identified that Dbp7 interacts with Dis3, an RNA exosome exoribonuclease subunit [62]. The RNA exosome is a conserved multi-subunit ribonuclease complex essential for degradation or processing of a variety of types of RNAs. To confirm the interaction, we performed Co-immunoprecipitation (Co-IP) using cells carrying Dbp7-TAP and Dis3-FLAG. Our Co-IP assays showed that Dbp7-TAP can pull down together with Dis3-FLAG, indicating that Dbp7 indeed interacts with Dis3 (Fig. 8A, Additional file 1: Fig. S4). To investigate the role of Dis3 in CENP-A centromere localization, we took advantage of a cold sensitive mutant, dis3-54. The dis3 mutant grows well at 33 °C, but is arrested at the restrictive temperature 20 °C. We found that Cnp1-GFP at centromeres in dis3-54 growing at room temperature for 8 h was almost undetectable (Fig. 8B and C), indicating that Dis3 is important for centromere loading. Consistent with the role of Dis3 in centromeres, a previous study showed that, similar to dbp7∆, silencing in the centromere cnt region is also lost in the dis3-54 mutant [63]. We also found that overexpression of Dis3 is unable to rescue the Cnp1-GFP distribution defect in dbp7∆ (Additional file 1: Fig. S5), and overexpression of Dbp7 also failed to restore centromeric localization of CENP-A in dis3-54 (Additional file 1: Fig. S6), suggesting that Dis3 and Dbp7 acts non-redundantly in regulation of CENP-A loading. Together, these data suggest that Dbp7 interacts together with Dis3 to promote the loading of CENP-A to centromeres.

Fig. 8.

Dbp7 and Dis3 interact to promote CENP-A loading through recruitment of CENP-A chaperone Sim3. A Co-IP showed that Dbp7 interacts with Dis3. Cell lysates from cells expressing Dbp7–TAP and Dis3-FLAG and from control cells expressing Dis3-FLAG only were subjected to immunoprecipitation with an antibody specific for TAP. Precipitated proteins were analyzed by immunoblotting using the indicated antibodies. n = 1. B Cnp1-GFP signal in the dis3-54 mutant is almost undetectable after incubation at room temperature for 8 h, indicating that Dis3 is required for CENP-A loading. Scale bars, 2 μm. C Quantification of Cnp1-GFP signal intensity in WT and the dis3-54 mutant cells. n = 40. Error bars represent mean and SD. ***, P < 0.001. D Dis3 physically associates with Sim3. Cell lysates from cells expressing Dis3-FLAG and Sim3-GFP and from control cells expressing Sim3-GFP only were subjected to immunoprecipitation with an antibody specific for FLAG. Precipitated proteins were analyzed by immunoblotting using the indicated antibodies. n = 1. E ChIP-qPCR assays showed that Dis3 is important for the association of Sim3 with centromeres. Indicated cells carrying Sim3-GFP were incubated at room temperature for 8 h. Immunoprecipitated DNAs were quantified by qPCR using primers specific for the cnt region. The data were normalized to the input. The Y-axis represents the fold change and the level of WT was set to 1. Data are shown as Mean ± SD (n = 3). **, P < 0.01

Dis3 is important for association of the CENP-A chaperone Sim3 with centromeres

Using TurboID-based proximity labeling coupled to mass spectrometry (PL-MS), a recent study showed that Dis3 also physically associates with the CENP-A chaperone Sim3 [64]. To verify if this is the case, we created the strain carrying Sim3-GFP and Dis3-FLAG, and conducted Co-IP assays using a FLAG antibody. Our Co-IP results confirmed the interaction between Sim3-GFP and Dis3-FLAG (Fig. 8D, Additional file 1: Fig. S7). To determine the role of Dis3 in the association of Sim3 with centromeres, we analyzed the dis3-54 mutant carrying Sim3-GFP and WT as a control by ChIP-qPCR using an anti-GFP antibody and primers against the centromere core region. These strains grew at room temperature for 8 h before collection for the ChIP-qPCR experiments. Our ChIP-qPCR results indicated that the level of Sim3-GFP is significantly reduced in the centromere core region in the dis3-54 mutant (Fig. 8E), indicating that Dis3 is important for recruitment of Sim3 to centromeres. Together, our data suggest that Dbp7 interacts with Dis3 mediate CENP-A loading to centromeres by recruiting the CENP-A chaperone Sim3.

Discussion

CENP-A serves as the epigenetic mark for specifying centromeres in most eukaryotes. How CENP-A is precisely positioned to centromere remains poorly understood. Studies in a variety of organisms have demonstrated that transcription occurs within CENP-A chromatin [23–25]. Growing evidence suggests non-coding RNAs generated from centromere transcription are involved in CENP-A loading to centromeres. Centromeric RNAs have been found to associate with CENP-C, a key inner kinetochore protein, in Drosophila, humans, and maize [65–67]. CENP-C, an essential factor for CENP-A assembly at centromeres, has a distinct RNA binding domain that interacts with centromeric RNAs [18–20, 66–68]. It has also been shown that non-coding centromeric RNAs associate with CENP-A [29, 69]. But the exact role of centromeric RNAs in CENP-A loading remains elusive.

In this study, we uncover a previously unrecognized mechanism for how RNA quality control factors promote CENP-A positioning at centromeres. Using a visual genetic screen, we identified that a conserved DEAD-box ATP-dependent RNA helicase, Dbp7, plays a critical role in CENP-A loading to centromeres. Both our imaging analysis and ChIP-seq data revealed that CENP-A in the dbp7∆ mutant is reduced at centromeres, and is mislocalized to non-centromeric regions. Consistent with the results, dbp7∆ cells exhibited irregular cell shape and chromosome missegregation. In addition, our reporter assays also showed that transcriptional silencing is lost in the centromeres of dbp7∆. DEAD box proteins typically bind RNAs and are involved in RNA-mediated processes. We found that Dbp7 interacts with centromere transcripts. Dbp7 contains an N-terminal disorder region, a middle DEAD/DEAH box helicase domain, and a C-terminal extension motif. Our domain-deletion analysis indicated that the middle DEAD/DEAH box helicase domain and the C-terminal extension motif are important for its centromere function. Importantly, the catalytic-dead Dbp7 point mutant is sufficient to lead to severe CENP-A mislocalization and chromosome segregation defects, indicating that its helicase activity is pivotal for its role in centromeres. These data are consistent with the idea that Dbp7 binds centromere RNAs to mediate CENP-A loading.

Consistent with a previous study [62], we found that Dbp7 interacts with the exosome subunit Dis3. Dis3 is a catalytic subunit of the exosome and plays a key role in the processing and degradation of RNA. Dis3 in S. pombe has been shown to be important for mitotic progression and pericentromeric heterochromatin formation [63]. Dis3 is also responsible for degradation of non-coding RNAs from centromere cores [38]. We showed in this study that Dis3 is crucial for CENP-A loading to centromeres. The association of CENP-A with centromeres is substantially lost in the dis3 mutant. A previous study also demonstrated that the silencing in centromere cores in the dis3 mutant is disrupted [63]. A recent proteomic work using PL-MS found that Dis3 interacts with the CENP-A chaperone Sim3 [64]. Sim3 is a homolog of the histone binding proteins NASP in humans. Sim3 in S. pombe directly binds Cnp1 and is required for loading of newly synthesized Cnp1 to centromeres [36]. The Arabidopsis homolog of Sim3 was also identified as a CENP-A chaperone [70]. In this study, we confirmed the interaction of Dis3 with Sim3, and further demonstrated that Dis3 is important for the recruitment of Sim3 to centromeres. Disruption of Dis3 results in loss of Sim3 from centromeres. In addition, our overexpression experiments suggest that Dis3 and Dbp7 acts non-redundantly in regulation of CENP-A loading. Nevertheless, we recognize the limitation of these overexpression studies. Future direct biochemical approaches, such as in vitro binding or reconstitution assays, will be important to understand how Dbp7, Dis3 and Sim3 work together to mediate CENP-A loading.

Taken together, our data suggest this model that Dbp7 and Dis3 acts together bind to centromeric RNAs, which in turn recruit Sim3 to mediate CENP-A loading to centromeres (Fig. 9). This mechanism is analogous to mechanisms used for heterochromatin assembly by which the exosome complex as well as RNAi machinery binds heterochromatin transcripts to recruit histone modifiers [46–48, 71–73].

Fig. 9.

Working model. Dbp7 and the exosome ribonuclease Dis3 act together to bind to centromeric RNAs, which in turn recruits the CENP-A chaperone Sim3 to mediate CENP-A loading to centromeres. The processing of centromere transcripts by Dbp7 and Dis3 also prevents accumulation of excessive R-loops in centromeres

R-loops, the three-stranded nucleic acid structures, can form during transcription. While they can be important for regulating gene expression and chromatin structure, excessive R-loops also pose a threat to genomic stability [33]. R-loops have been identified at centromeres in various organisms [34, 74–76]. Accumulation of abundant R-loops at centromeres in budding yeast contributes to defects in kinetochore function and chromosomal instability [34]. It has also been reported that higher levels of R-loops at centromeres in plants correlates with a reduced localization of CENP-A homolog, CENH3 [77]. Basrai’s group recently showed that an RNA-binding protein Hpr1 plays an important role in preventing the accumulation of R-loops at centromeres in budding yeast [34]. The exact mechanism of how R loops are resolved at centromeres remains unknown. Rnh1 is an RNase H enzyme important for R-loop resolution in fission yeast [50]. Here we showed that deletion of rnh1⁺ results in loss of centromere silencing and decrease of CENP-A in centromeres, consistent with the idea that excessive R-loops disrupt centromere integrity. In addition, we found that overexpression of Rnh1 also causes significant reduction of CENP-A and loss of silencing in centromeres. These data support the notion that maintaining proper homeostasis of R-loops in centromeres is important for CENP-A loading.

It has been reported that DEAD-box helicases and the RNA exosome complex participate in R loop resolution [61, 78–80]. Our study here showed that deletion of dbp7⁺ leads to accumulation of R-loops at centromeres, indicating that Dbp7 prevents excessive centromeric R-loops. We further showed that its helicase activity is required for the role of Dbp7 in R-loop resolution. In addition, Dbp7 binds to centromeric RNAs and physically associates with the exosome ribonuclease Dis3. Our data suggest that Dbp7 and Dis3 work together to remove redundant R-loops at centromeres (Fig. 9). Together, our results reveal a mechanism underlying R loop resolution at centromeres. Nevertheless, the CENP-A mislocalization and silencing defects in centromeres in dbp7∆ is significantly stronger than the rnh1∆, reinforcing the notion that Dbp7 acts an additional role in direct recruitment of CENP-A by interacting with Dis3 (Fig. 9).

Dbp7 is a DEAD-box family protein conserved across eukaryotes, but it remains poorly studied [81]. It will be interesting for the future to further investigate whether a similar mechanism underlying Dbp7-mediated centromere assembly is shared across different organisms. Both human homologs of Dbp7 and Dis3 are associated with multiple types of human cancers [58, 59, 82]. How Dbp7 and Dis3 can mechanistically cause cancer remains elusive. It will be also interesting to know whether CENP-A loading and centromeric R loop resolution mediated by these RNA quality control factors contributes to the origin of these cancers.

Conclusions

This study thus reveals a novel role of centromeric RNAs in CENP-A loading and highlights the importance of RNA quality control factors in CENP-A chromatin assembly and genome integrity at centromeres.

Methods

Strains, media and genetic analysis

The fission yeast strains are listed in Additional file 1: Table S1. The standard media and genetic experiments were employed as described previously [83]. Yeast extract with supplements (YES) was used as a complete culture and pombe glutamate medium (PMG) was used as a minimum media. For growth assays, a series of tenfold dilutions with a starting concentration of 1 × 10⁷ cells/ml were spotted on the designated plates with indicated drugs and incubated at 30 ºC for 3–4 days. The minichromosome loss assays were performed as described previously [84]. The epitope-tagged alleles and domain-deletion mutants were constructed at the endogenous loci by homologous recombination method and confirmed by PCR and sequencing.

ChIP

Chromatin immunoprecipitation (ChIP) was conducted as described previously [52]. Briefly, 50 ml of yeast cells were grown to log phase at 30 ºC and cross-linked with formaldehyde (1% final concentration) for 30 min at RT. For ChIP analysis of the dis3-54 mutant, cells grew at room temperature for 8 h before crosslinking. After treating with Zymolase at RT for 30 min, cells were collected and sonicated by an Ultrasonic Processor. 60 μl of the solubilized chromatin was used as input and 600 μl was used for IP with IgG sepharose (GE Healthcare, 17096901), anti-acetyl-histone H3 (Lys9) antibody (Millipore 2658782), or anti-GFP antibody (Abcam, ab290). Immunoprecipitated DNA was purified using NucleoSpin® Gel and PCR Clean-Up kit (Takara Bio, USA). qPCR was performed with Fast SYBR Green Master Mix (Applied Biosystems) on a Bio-Rad iCycler and analyzed with iCycler iQ Optical System Software. Relative enrichments were calculated as the ratio of product of interest to control (act1⁺) in IP over input. 1 μl of ChIP or whole cell extract samples were calculated following the standard ΔΔCt method. The primers for qPCR are listed in Additional file 1: Table S2.

ChIP-seq

ChIP-seq was performed as described previously [52]. Briefly, after ChIP experiments as performed above using an anti-GFP antibody (Abcam, ab290), libraries were prepared and then sequenced by Illumina HiSeq2500 platform to generate single end 50 nt reads. Read alignments to the S. pombe genome (ASM294v3/GCF_000002945.2) were performed using Bowtie2 version 2.4.4. [85]. BigWig files for genome coverage and log twofold change of Cnp1-GFP signal between the dbp7∆ mutant and wild type were generated using deepTools version 3.5.0 with a bin size of 10. Reads were normalized to reads per kilobase per million (RPKM) [86]. All data was visualized in the integrative genome viewer (IGV) [87].

RIP

RNA immunoprecipitation (RIP) was carried out as described previously [88]. Briefly, 100 ml of yeast cells were grown to mid-log phase at 30 ºC and then incubated at restrictive 36 °C for 8 h before cross-linking with formaldehyde (1% final concentration) for 20 min at RT. After treating with Zymolase at RT for 30 min, cells were collected in FA lysis buffer and sonicated by an Ultrasonic Processor. Following RNase-free DNase I treatment at 37 ºC for 15 min, cell extracts were incubated with 30 μl IgG sepharose (GE Healthcare, 17096901) or 1 μl anti-GFP antibody (Abcam, ab290) for 1 h. Immunoprecipitated RNAs were purified using NucleoSpin® RNA kit (Takara Bio, USA). The Input and IP samples were transcribed by Takata RT-PCR kit and analyzed by qPCR. qPCR was performed with Fast SYBR Green Master Mix (Applied Biosystems) on a Bio-Rad iCycler and analyzed with iCycler iQ Optical System Software. Relative enrichments were calculated as the ratio of product of interest to control (act1⁺) in IP over input, and the RNA level immunoprecipitated from untagged strain by RIP was set to 1. 1 μl of RIP or whole cell extract samples were calculated following the standard ΔΔCt method. The primers for qPCR are listed in Additional file 1: Table S2.

DRIP

DNA-RNA Immunoprecipitation (DRIP) was conducted as described [89]. Briefly, 100 ml of yeast cells were grown to log phase at 30 ºC and collected by centrifuge for 2 min at 3000 rpm. The genomic DNA was extracted using DNeasy Blood & Tissue Kit (Qiagen) and sheared by an Ultrasonic Processor. The DNA–RNA hybrids were precipitated with antibody S9.6 (Millipore MABE1095). Immuoprecipitated DNA was recovered with Chelex-100. qPCR was performed with Fast SYBR Green Master Mix (Applied Biosystems) on a Bio-Rad iCycler and analyzed with iCycler iQ Optical System Software. Relative enrichments were calculated as the ratio of product of interest to control (act1⁺) in IP over input. 1 μl of DRIP or whole cell extract samples were calculated following the standard ΔΔCt method. The primers for qPCR are listed in Additional file 1: Table S2.

Co-IP

Co-immunoprecipitation (Co-IP) was conducted as described previously [90]. Briefly, 100 ml of logarithmic phase cells were collected and resuspended in 100 μl lysis buffer (20 mM HEPES at pH 7.5, 5 mM EDTA, 100 mM NaCl, 0.1% Nonidet P-40) with 1 × protease inhibitors (Sigma, P8215), 1 μM PMSP, and 1 mM DTT, and broken by bead beating at 4 ºC. Lysates were incubated with IgG sepharose (GE Healthcare, 17,096,901) or anti-FLAG antibody (Sigma SLBS3530V) at 4 ºC for 2 h. The Input and IP samples were eluted in SDS loading buffer and analyzed by Western blot using anti-GFP (Abcam, ab290), peroxidase anti-peroxidase (Sigma, P1291), or anti-FLAG (Sigma SLBS3530V) antibody.

Microscopy

Live or DAPI-stained fixed cells were imaged using the Delta Vision System (Applied Precision, Issaquah, WA). Images were taken as z-stacks of 0.2-mm increments with an oil immersion objective (× 100), and deconvolved using SoftWoRX2.50 software (Applied Precision). The Cnp1-GFP signals were quantified by using ImageJ 1.54 g (NIH, USA). Immunostaining assays are described as previously [83]. Briefly, 20 ml of exponentially growing cells were collected by centrifugation and subsequently fixed with −20 °C methanol. Dbp7-TAP and Dbp7-S-tag were detected with the monoclonal antibody against S-peptide (Invitrogen, MA1-981) followed by Alexa Fluor 488 anti-mouse secondary antibody (Themo Fisher, R37120). Nuc1-GFP was stained with anti-GFP primary antibody (Abcam, ab290) followed by Alexa Fluor 594 secondary antibody (Themo Fisher, A-11012). Slide preparation was finished by adding mounting medium containing DAPI.

Authors’ contributions

F.L. conceived and supervised the project. J.G., F.G., Q.D., Z.L., L.M., M.A., J.L., and J.Y. performed experiments and data analysis. F.L., and J.G. wrote the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the NIH grant (2R35GM134920 to F.L.).

Data availability

The ChIP-seq data reported in the paper have been deposited to the GEO database under accession code: GSE285031 [91]. The image data have been deposited at Figshare (DOI: https://doi.org/10.6084/m9.figshare.32393697) [92].

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.