Pab2 is necessary for proper growth and meiosis

Given the importance of polyadenylation and poly(A)-binding proteins in various cellular transactions, we first constructed a pab2 (SJAG_04601) deletion strain (pab2Δ), in the fission yeast Schizosaccharomyces japonicus and characterized the pab2∆ cells. The pab2Δ cells exhibited slow growth, particularly pronounced at lower temperatures (18°C and 26°C) compared to wild-type (WT) cells (S1A and S1B Fig). This observation suggests that S. japonicus Pab2 has a critical role in growth, which may not be evident in S. pombe, where pab2 deletion does not significantly hinder growth at 30°C [32]. In addition, microscopic analyses demonstrated that approximately 10% of pab2Δ cells displayed an abnormal DAPI staining pattern (S1C Fig), suggesting that Pab2 is required for normal mitotic cell cycle progression. Moreover, the number of asci produced by homothallic pab2Δ cells was reduced compared to that of homothallic WT cells when nitrogen sources were depleted (S1D Fig), and the mating efficiency was significantly decreased in pab2∆ cells (S1E Fig). These results indicate that Pab2 is required for efficient mating in S. japonicus, while Pab2 is dispensable for efficient mating in S. pombe [33]. Our findings demonstrate that Pab2 is essential for proper growth and meiosis in S. japonicus, revealing a novel role for this poly(A)-binding protein in these fundamental cellular processes and highlighting a potential divergence between the two fission yeast species.

The nuclear poly(A)-binding protein Pab2 prevents the untimely accumulation of putative meiotic genes in S. japonicus

Pab2 and its orthologs play a role in RNA metabolism [16,34–36]. To investigate the function(s) of Pab2 in mitosis and meiosis, transcriptome analysis of pab2Δ was performed using two rounds of RNA sequencing. The RNA-seq results were reproducible, with highly significant overlaps showing more than a 2-fold increase and less than a 0.5-fold decrease in pab2Δ between two independent experiments (S1F Fig). We identified a total of 389 and 83 transcripts that were upregulated and downregulated in pab2Δ, respectively, across both biological replicates (S1 and S2 Tables). Gene ontology (GO) analysis revealed that 22 out of 389 of the Pab2-suppressed genes are meiosis-related (p < 10^-3.28) (Fig 1A and S3 Table), indicating that Pab2 plays a role in repressing potential meiosis-related gene expression. Consistent with previous findings in S. pombe showing that Pab2 suppresses untimely expression of meiotic genes [33,36], RT-qPCR analysis confirmed elevated mRNA levels for four examined putative meiotic genes in pab2Δ (Fig 1B). In contrast, GO analysis of the downregulated genes suggested that Pab2 promotes the expression of genes involved in small molecule (amino acids) metabolic/catabolic processes (S4 Table), which likely contributes to normal vegetative growth of S. japonicus (S1A and S1B Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Putative meiotic mRNA accumulation in S. japonicus pab2

Δ. (A) Expression of putative meiosis-related genes in wild-type (WT) and the pab2Δ cells. RNA-seq data of four representative meiotic genes (top) and four other highly expressed meiotic genes (bottom) are shown. (B) RT-qPCR data from four putative meiotic genes in WT and pab2Δ cells. The four putative meiotic mRNAs (mei4, ssm4, rec8, and spo5) were normalized to act1 mRNA to determine their relative expression levels. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05, **p < 0.01, and ***p < 0.005).

https://doi.org/10.1371/journal.pgen.1011647.g001

We previously reported that the Zn-finger protein Red1 and its associated proteins promote selective degradation of meiotic mRNAs during vegetative growth in S. pombe [16,37–39]. As anticipated, the levels of three putative meiotic mRNAs (mei4, rec8, and spo5) were accumulated in S. japonicus red1Δ cells (S1G Fig). Moreover, Pab2 colocalizes with Red1 in nuclear foci (S1H Fig), as reported in S. pombe [39,40], suggesting that Pab2 and Red1 nuclear foci are important primarily for putative meiotic mRNA degradation. These data indicate a conserved role for Pab2 and Red1 in selective mRNA degradation in both S. japonicus and S. pombe.

Pab2 participates in the suppression of transcripts derived from constitutive heterochromatin

We identified that transcripts from centromeric and telomeric domains were upregulated in pab2Δ (Fig 2A and S1 Table). Transcripts with elevated levels included retrotransposon fragments within constitutive heterochromatin and three genes embedded within the mating-type locus (SJAG_05153, SJAG_05154, and SJAG_01113). This finding was unexpected as, to the best of our knowledge, Pab2 has not previously been directly implicated in the formation of constative heterochromatin in either S. japonicus or S. pombe. Consistent with our RNA-seq data, RT-qPCR confirmed that the transcript levels from centromeric and telomeric regions (S2A Fig) were significantly elevated in pab2Δ (Fig 2B). Because transcripts from repetitive sequences are critical for RNA-based heterochromatin assembly in fission yeast [24,41,42], we hypothesized that heterochromatin assembly was adversely affected in S. japonicus lacking Pab2.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Pab2 contributes to suppressing transcripts derived from constitutive heterochromatin.

(A) RNA-seq read coverage of transcripts derived from centromeres, telomeres, and the mating-type locus (located within the pericentromeric heterochromatin of chromosome III) in wild-type (WT, top panels), pab2Δ (middle panels), and clr4–1 (bottom panels) strains. Note that the clr4–1 panels use a different scale for normalized reads (RPM). (B) RT-qPCR data from centromeric and telomeric transcripts in WT, pab2Δ, and clr4–1 strains. Centromeric (cen) and telomeric transcripts (tel) were normalized to act1 mRNA to determine their relative expression levels. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05 and ***p < 0.005). (C) RT-qPCR data from mating-type transcripts in WT, pab2Δ, clr4–1, and red1Δ strains. Transcripts from Mc-like and Pc-like genes (Mc-like and Pc-like) were normalized to act1 mRNA to determine their relative expression levels. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (**p < 0.01, ***p < 0.001, and ****p < 0.0001; N.S.: not significant). (D) RNA immunoprecipitation (RNA-IP) using anti-GFP antibody in Pab2-GFP and untagged parental (No tag) strains. Precipitated RNAs were subjected to RT-qPCR. Centromeric RNA (cen) and five other mRNAs (mei4, ssm4, rec8, tbp1, and cdc2) were normalized to act1 mRNA to determine their relative enrichment. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05 and **p < 0.01; N.S.: not significant).

https://doi.org/10.1371/journal.pgen.1011647.g002

To investigate heterochromatin in S. japonicus, we first isolated clr4–1, a temperature-sensitive mutant of the histone H3 Lys9 methyltransferase Clr4/SUV39 (S2B Fig), as a control strain since Clr4 is essential for S. japonicus viability [24]. Genome sequencing results indicated that the clr4–1 mutant carried two nucleotide substitutions: one leading to an amino acid substitution (TGT to CGT, Cys344Arg) in the SET domain responsible for histone H3 Lys9 methyltransferase activity, and the other being a silent mutation (ACA to ACG, Thr51) in the chromodomain (S2C Fig). Additionally, RNA-seq data demonstrated that the levels of both centromeric (SJATN_00017) and telomeric (SJAG_03765 and SJAG_03766) transcripts increased substantially in clr4–1 (Fig 2A and S5 Table), and the increase was confirmed by RT-qPCR (Fig 2B). GO analysis indicated that transcription levels of genes involved in the amino acid metabolic process were reduced in clr4–1 (S6 Tables).

Interestingly, mRNAs derived from three genes encoding Mc-like (SJAG_05153), Pc-like (SJAG_01113), and Pi (SJAG_05154) proteins within the mating-type locus were not accumulated in clr4–1 (Fig 2A), despite their location within pericentromeric heterochromatin of chromosome III [21]. We hypothesized that these mRNAs are degraded by the Pab2/Red1-dependent RNA elimination pathway due to their association with meiosis. RT-qPCR results revealed that mRNAs encoding Mc-like and Pc-like proteins were upregulated in both pab2∆ and red1∆, but not in clr4–1 (Fig 2C), indicating that the genes within the mating-type locus are primarily repressed by the RNA elimination mechanism in S. japonicus.

Our RNA-seq data also demonstrated that Pab2- and Clr4-target genes significantly overlapped (S2D Fig and S7 and S8 Tables), suggesting that Pab2 and Clr4 cooperated to suppress and promote the expression of specific genes. GO analysis of the downregulated genes in both pab2∆ and clr4–1 mutant strains showed enrichment in gamma-aminobutyric acid metabolic process, small molecule catabolic process, gamma-aminobutyric acid catabolic process, non-proteinogenic amino acid metabolic process, and arginine biosynthetic process via ornithine (S2E Fig). On the other hand, no significant enrichment was found in GO analysis of the upregulated genes in both pab2∆ and clr4–1 mutant strains.

A previous study demonstrated that non-coding RNAs derived from centromeres are transcribed by RNA polymerase II (Pol II) in S. pombe [43], prompted us to examine Pol II occupancy at centromeres in pab2∆. ChIP-qPCR analysis revealed no significant increase in Pol II occupancy in pab2∆ (S2F Fig). This is consistent with the modest upregulation of heterochromatic transcripts observed in pab2∆ (Fig 2A and 2B), particularly when compared to the substantial increase in both Pol II occupancy (S2F Fig) and centromeric transcript levels (Fig 2A and 2B) observed in clr4–1. These findings suggest that transcriptional gene silencing at heterochromatin is largely maintained even in pab2∆.

Given that Pab2 is an RNA-binding protein, we assumed that it binds to transcripts derived from constitutive heterochromatin and putative meiotic genes. To examine this assumption, we performed RNA immunoprecipitation (RNA-IP) followed by RT-qPCR. As shown in Fig 2D, Pab2 preferentially bound to RNAs derived from centromeres and putative meiotic mRNAs but did not preferentially associate with tbp1 and cdc2 mRNAs, which encode the TATA-binding protein Tbp1/TBP and the cyclin-dependent kinase Cdc2/Cdk1, respectively. These findings suggest that Pab2 selectively interacts with certain RNAs.

Pab2 facilitates H3K9 methylation at constitutive heterochromatin

One of the hallmarks of heterochromatin is histone H3 Lys9 di- and tri-methylation (H3K9me2/H3K9me3), which provides binding sites for chromodomain proteins such as HP1 [4,5]. While both H3K9me2 and H3K9me3 are repressive marks that are bound by chromodomain proteins, H3K9me3 is generally associated with a stronger and more stable form of transcriptional repression, often leading to heterochromatin formation, compared to H3K9me2 [44]. Since heterochromatic silencing is partially disrupted in pab2∆, we conducted chromatin immunoprecipitation followed by qPCR (ChIP-qPCR) to assess H3K9me levels in pab2Δ. ChIP-qPCR results revealed that H3K9me2 at centromeres and telomeres was significantly decreased in pab2Δ and clr4–1 (Fig 3A). In contrast, H3K9me3 was significantly reduced at centromeres but not at telomeres in pab2∆ and clr4–1 (Fig 3A), suggesting the more robust or redundant mechanism of heterochromatin assembly at telomeres. These results suggest that Pab2 is not absolutely required for H3K9me, but it promotes proper H3K9me in S. japonicus.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Pab2 plays an important role in constitutive heterochromatin.

(A) H3K9me2, H3K9me3, and total H3 enrichment at centromeres and telomeres in wild-type (WT), pab2Δ, and clr4–1 cells were assessed by ChIP-qPCR. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; N.S.: not significant). (B) No noticeable heterochromatin islands at the mei4⁺ and ssm4⁺ loci in S. japonicus. ChIP-qPCR analysis of H3K9me2 and total H3 enrichment at these loci and centromeres in WT, pab2∆, and clr4–1 was conducted. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (***p < 0.001 and ****p < 0.0001; N.S.: not significant). (C) ChIP-qPCR analysis of H3K9me2 and total H3 enrichment at centromeres in WT, pab2∆, and red1∆. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (**p < 0.01; N.S.: not significant). (D) Thiabendazole (TBZ) sensitivity assay of WT, pab2Δ, and clr4–1 cells. Ten-fold serial dilutions were spotted onto complete medium plates in the presence or absence of TBZ and grown at 30°C.

https://doi.org/10.1371/journal.pgen.1011647.g003

As described above, S. pombe has facultative heterochromatin at various loci, with two representative heterochromatin islands found at two meiotic gene loci, mei4⁺ and ssm4⁺ [7,45–47]. Because the genomic regions surrounding these loci are syntenic between S. japonicus and S. pombe, we reasoned that S. japonicus might also have heterochromatin islands at the orthologous loci. However, we did not observe any significant enrichment of H3K9me2 at either locus by ChIP-qPCR (Fig 3B), suggesting that the assembly of facultative heterochromatin at meiotic genes is not conserved even among Schizosaccharomyces species. Note that facultative heterochromatin at the meiotic genes is not always observed even in WT cells of S. pombe [48]. Besides, considering that both Pab2 and Red1 suppressed inferred meiotic mRNAs during vegetative growth (Figs 1 and S1G), we propose that Pab2, but not Red1, has two separable functions: mRNA degradation and heterochromatin assembly.

Previous studies reported that RNA degradation proteins, including Red1, Mmi1, and Erh1, were crucial for the assembly of facultative but not constitutive heterochromatin in S. pombe [17,31,37,45,47,49]. Similarly, the levels of H3K9me2 at centromeres in red1∆ were comparable to those in WT (Fig 3C), indicating that the Red1-dependent RNA degradation pathway is dispensable for the formation of constitutive heterochromatin in S. japonicus.

The expected increase in centromeric transcripts was not observed in pab2Δ (Fig 2B), although H3K9me2/H3K9me3 were significantly decreased (Fig 3A). We hypothesized that Pab2-independent pathways might degrade the centromeric transcripts. Indeed, centromeric transcripts were also elevated in cid14∆, S. japonicus lacking Cid14/TENT4 — a constituent of the nuclear exosome-targeting complex TRAMP [50], but the increase was more pronounced in cid14Δ (S3A Fig). This result indicates that TRAMP-dependent RNA decay suppresses centromeric transcripts at the post-transcription level, as described in S. pombe [50]. Additionally, combining the pab2^RRM3A mutation (see below) with cid14∆ resulted in a significant reduction of H3K9me2 (S3B Fig). Thus, it is plausible that centromeric transcripts upregulated in pab2∆ are degraded by the TRAMP-mediated RNA elimination system, thereby masking the impact of pab2 deletion.

It has been demonstrated that centromeric heterochromatin is vital for proper chromosome segregation, and loss of heterochromatin resulted in sensitivity to thiabendazole (TBZ), a microtubule-destabilizing agent [51–53]. If the partial loss of H3K9me2/H3K9me3 is biologically significant, we would expect TBZ sensitivity to be observed in pab2∆ cells. Indeed, we found that both pab2Δ and clr4–1 exhibited sensitivity to TBZ (Fig 3D). From these results, we proposed that Pab2 is essential for the assembly of constitutive heterochromatin in S. japonicus.

The Pab2 intrinsically disordered region, which coordinates phase separation, is required for heterochromatin assembly

Intrinsically disordered regions (IDRs) are essential for forming biomolecular condensates via phase separation [54,55]. During vegetative growth in S. japonicus, Pab2 predominantly forms a single nuclear focus (S1H and S4A Figs), consistent with observations in S. pombe [33,40], suggesting that Pab2 possesses an IDR that drives condensate formation. Indeed, DISOPRED3 [56] predicted that the C-terminal domain of Pab2 (amino acids: 136–174), rich in arginine and glycine residues, is an intrinsically disordered region (Fig 4A). Treatment with 1,6-hexanediol (1,6-HD), which disrupted membrane-less biological condensates [57,58], abolished Pab2 foci, and these foci reappeared upon 1,6-HD removal (S4B Fig). As 1,6-HD treatment did not affect steady-state Pab2-GFP levels (S4C Fig), these results indicate that Pab2 forms phase-separated nuclear condensates in a C-terminal IDR-dependent manner.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. The Pab2 intrinsically disordered region (IDR) is required for heterochromatin assembly and condensate formation.

(A) Prediction of Pab2 IDR using DISOPRED3. The C-terminal region of Pab2 (amino acids: 136–174) is predicted to be intrinsically disordered. The N-terminal domain (N) is predicted to be partially disordered, while the RNA-recognition motif (RRM) is predicted to be ordered. (B) Western blot analysis of Pab2-GFP and Pab2ΔIDR-GFP protein levels. Pcn1/PCNA serves as a loading control. (C) Fluorescent microscopy of Pab2-GFP (WT) and Pab2∆IDR-GFP (∆IDR) localization. The white dotted lines indicate cell boundaries. Scale bars, 2 μm. (D) ChIP-qPCR analysis of H3K9me2 and total H3 enrichment at centromeres and telomeres. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (**p < 0.01; N.S.: not significant). (E) Schematic representation and amino acid sequences of the IDR of Pab2 and Pab2^RA mutant proteins. (F) in vitro droplet formation assay using purified Pab2 and Pab2^RA proteins. Scale bars, 5 μm. (G) Western blot analysis of Pab2-GFP (WT) and Pab2^RA-GFP (RA). Pcn1/PCNA serves as a loading control. (H) Fluorescent microscopy of Pab2-GFP (WT) and Pab2^RA-GFP (RA) localization. The white dotted lines indicate cell boundaries. Scale bars, 2 μm. (I) ChIP-qPCR analyses of H3K9me2 and total H3 enrichment at centromeres and telomeres. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (**p < 0.01; N.S.: not significant). (J) ChIP-qPCR analyses of 3 × FLAG-Clr4 and Mit1-GFP levels at centromeres and telomeres in WT and pab2∆ strains. An untagged strain (No tag) was used as a negative control. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05). (K) ChIP-qPCR analysis of H3K9me2 and total H3 enrichment at centromeres and telomeres in WT, mit1∆, and clr4–1. Data represent mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (***p < 0.001 and ****p < 0.0001; N.S.: not significant).

https://doi.org/10.1371/journal.pgen.1011647.g004

Given the roles of phase separation in chromatin dynamics [59–61], we investigated whether Pab2 condensates contribute to heterochromatin assembly. To this end, we generated a strain expressing Pab2ΔIDR-GFP, a GFP-tagged Pab2 protein lacking the C-terminal IDR (S4D Fig). Western blotting confirmed comparable expression levels of Pab2-GFP and Pab2ΔIDR-GFP (Fig 4B). However, microscopic analysis revealed that Pab2ΔIDR-GFP failed to either localize to the nucleus or form nuclear foci (Fig 4C), indicating that the IDR is essential for proper Pab2 localization. ChIP-qPCR demonstrated reduced H3K9me2 levels at centromeres and telomeres in the Pab2ΔIDR-GFP strain (Fig 4D). Consistent with the partial disruption of constitutive heterochromatin, Pab2ΔIDR-GFP cells were more sensitive to TBZ than WT cells (S4E Fig). Moreover, we observed significantly increased levels of putative meiotic mRNAs, but not centromeric or telomeric transcripts, in Pab2∆IDR-GFP cells (S4F Fig). These observations suggest that the Pab2 C-terminal domain is crucial for both selective mRNA decay and heterochromatin assembly. However, the data also suggest that TRAMP-dependent RNA decay pathway can compensate for the loss of Pab2 in degrading heterochromatic transcripts.

Arginine (Arg) residues facilitate interactions among disordered domains via charge-charge interactions and cation-π interactions [62], and these interactions stabilize protein/RNA condensates through the multivalency of the guanidinium group [63]. We therefore investigated whether Arg residues within the Pab2 IDR (Fig 4E) are crucial for condensate formation. We expressed and purified the MBP (maltose-binding protein)-Pab2–6 × His recombinant protein in E. coli (S4G and S4H Fig). Under our experimental conditions, recombinant Pab2 formed spherical structures at 2.5 µM (Fig 4F) that could fuse with each other (S1 Movie), demonstrating its ability to form liquid-like droplets in vitro. We also purified recombinant Pab2^RA, in which all Arg residues in the IDR were replaced with Ala (Fig 4E). Recombinant Pab2^RA protein formed spherical structures only at significantly higher concentration (25 µM) compared to wild-type Pab2 (Fig 4F), indicating that the Arg residues are critical for efficient condensate formation in vitro.

To assess the role of these Arg residues in vivo, we constructed a strain expressing Pab2^RA-GFP. This mutation did not adversely affect steady-state Pab2-GFP protein levels (Fig 4G). However, Pab2^RA-GFP failed to form nuclear foci, exhibiting diffuse nuclear localization instead (Fig 4H). This confirms that Arg residues in the IDR are indispensable for Pab2 condensate assembly in vivo. In addition, Pab2^RA-GFP cells exhibited significantly decreased H3K9me2 levels at centromeres and telomeres (Fig 4I) and elevated sensitivity to TBZ than Pab2-GFP cells (S4E Fig). Moreover, putative meiotic mRNA levels increased, while heterochromatin transcripts remained unchanged in Pab2^RA-GFP cells (S4I Fig), mirroring the observations in Pab2∆IDR-GFP cells. These results underscore the crucial role of the IDR in heterochromatin assembly.

Given that biomolecular condensates can increase the local concentration of specific factors [64–66], we examined whether Pab2 condensates recruit or stabilize the association of heterochromatin factors, such as Clr4, at heterochromatin. ChIP-qPCR analysis revealed significantly reduced Clr4 binding to constitutive heterochromatin in pab2∆ (Fig 4J). We also observed decreased levels of Mit1 (SJAG_03982), a Mi-2-like chromatin remodeler in SHREC [15,67], at centromeres and telomeres in pab2∆ (Fig 4J). The importance of Mit1 in heterochromatin assembly was supported by the substantial loss of H3K9me2 and increased TBZ sensitivity in mit1∆ cells (Figs 4K and S4J). These findings suggest that Pab2 promotes heterochromatin assembly by facilitating the recruitment of Clr4 and Mit1 to constitutive heterochromatin through its condensates.

Pab2 RNA-binding activity is critical for condensate formation and heterochromatin assembly

While we observed decreased H3K9me enrichment at centromeres and telomeres of pab2Δ (Fig 3A), it remained unclear whether the RNA-binding activity of Pab2 is necessary for heterochromatin assembly. To investigate this, we constructed three mutant strains: Pab2ΔRRM, expressing Pab2 lacking the RNA recognition motif (RRM); Pab2^RRM3A, carrying the substitutions of three aromatic residues (Y61A, F101A, and Y103A) critical for RRM activity [68]; and Pab2^RRM3A-RA, possessing both the RRM mutations and multiple Arg to Ala substitutions in the IDR (Fig 5A). Western blot analysis demonstrated that the steady-state levels of Pab2∆RRM, Pab2^RRM3A, and Pab2^RRM3A-RA were comparable to that of Pab2 (Fig 5B). However, we did not readily observe nuclear foci in cells expressing any of these mutant proteins (Fig 5C), indicating that the formation of Pab2 foci depends on RRM. To examine whether each Pab2 mutant protein binds to centromeric transcripts, we performed RNA-IP. As shown in Fig 5D, Pab2 and Pab2^RA preferentially bound to centromeric transcripts, whereas Pab2∆RRM, Pab2^RRM3A, and Pab2^RRM3A-RA lost their ability to associate with centromeric RNAs. These observations highlight the importance of the RRM for Pab2 RNA-binding and for Pab2 nuclear condensate formation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. The Pab2 RNA-recognition motif (RRM) is required for heterochromatin assembly.

(A) Schematic representation of Pab2 constructs: Pab2ΔRRM lacks the RRM; Pab2^RRM3A contains three amino acids substitutions (Y61A, F101A, and Y103A); Pab2^RRM3A-RA combines the RRM mutations of Pab2^RRM3A with additional arginine-to-alanine mutations in the IDR. (B) Western blot analysis of Pab2-GFP (WT), Pab2ΔRRM-GFP (ΔRRM), Pab2^RRM3A-GFP (RRM3A), and Pab2^RRM3A-RA-GFP (RRM3A-RA) with Pcn1/PCNA as a loading control. (C) Fluorescence microscopy images of Pab2-GFP and its RRM mutants. The white dotted lines indicate cell shapes. Scale bars, 2 μm. (D) RNA immunoprecipitation (RNA-IP) followed by RT-qPCR to assess RNA-binding activity in cells expressing untagged Pab2 (No tag), Pab2-GFP, Pab2ΔRRM-GFP, Pab2^RRM3A-GFP, Pab2^RA-GFP, or Pab2^RRM3A-RA-GFP. Centromeric RNA (cen) was normalized to act1 mRNA to determine their relative enrichment. Data are presented as mean ± SD from three independent trials. (Left) Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with ‘Pab2’ as the reference sample (**p < 0.01; N.S.: not significant). (Right) Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05). (E) ChIP-qPCR analysis of H3K9me2 and total H3 enrichment at centromeres using the indicated strains. Data are presented as mean ± SD from three independent experiments. (Left and middle) Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (**p < 0.01, ***p < 0.001 and ****p < 0.0001; N.S.: not significant). (Right) Statistical significance was determined using a two-tailed unpaired t-test (***p < 0.005; N.S.: not significant). (F) ChIP-qPCR analysis of Pab2 and its mutant proteins at centromeres. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with ‘No tag’ as the reference sample (*p < 0.05; N.S.: not significant).

https://doi.org/10.1371/journal.pgen.1011647.g005

We then conducted RT-qPCR to assess RNA expression in the Pab2 mutant strains. We found increased transcript levels in the following strains: pab2ΔRRM showed elevated levels of transcripts from mei4⁺, ssm4⁺, spo5⁺, rec8⁺, and centromeres; pab2^RRM3A exhibited increases in transcripts derived from ssm4⁺and rec8⁺; and in pab2^RRM3A-RA had elevated levels of transcripts from mei4⁺, ssm4⁺, spo5^+, rec8⁺, and centromeres (S5A Fig). Additionally, H3K9me2 ChIP-qPCR revealed reduced levels of H3K9me2 at centromeres in all the Pab2 mutant strains, as well as in pab2∆ cells (Fig 5E). Consistent with this observation, all Pab2 mutant strains exhibited increased sensitivity to TBZ compared to WT cells (S5B Fig), indicating that the loss of heterochromatin in these mutants is biologically significant. Thus, we conclude that the RRM of Pab2 is critical for its functions in heterochromatin assembly and silencing of putative meiotic genes.

Pab2 preferentially associates with centromeres in both RRM and IDR-dependent manners

Pab2 promotes H3K9 methylation at constitutive heterochromatin and forms nuclear condensates, suggesting that Pab2 directly associates with or is localized in proximity to heterochromatin. To initially investigate this potential association, we examined the colocalization of Pab2 with known heterochromatin markers. Fluorescent microscopy demonstrated the minimal colocalization of Pab2 with Chp1 (S5C Fig), a chromodomain protein that binds to H3K9me [69]. Considering this result with the earlier observation that Pab2 foci colocalize with Red1 (S1H Fig), a protein dispensable for heterochromatin formation (Fig 3C), these results suggest that the observed Pab2 nuclear foci are likely involved in selective RNA degradation, rather than in heterochromatin assembly.

To directly assess Pab2’s association with heterochromatin, particularly at centromeres, we performed ChIP-qPCR analysis using Pab2-GFP fusion proteins. This analysis revealed a preferential interaction of Pab2 at centromeres (Fig 5F), providing strong evidence for its direct association with this heterochromatic region. Importantly, analysis of Pab2 mutants revealed that while Pab2∆RRM, Pab2^RRM3A, and Pab2^RA mutant proteins retained centromere binding, the Pab2^RRM3A-RA mutant significantly lost this interaction (Fig 5F). Taken together, these results demonstrate that Pab2 preferentially associates with centromeres in both RRM- and IDR-dependent manners, as only simultaneous disruption of both domains abolishes binding.

Red5 and Rmn1, which interact with Pab2, are essential for constitutive heterochromatin assembly

To determine how Pab2 facilitates constitutive heterochromatin formation, we sought to identify Pab2-binding proteins using TurboID, a proximity-dependent protein labeling method [70]. Mass spectrometry analyses of biotinylated protein-enriched fractions revealed several proteins specifically enriched in the Pab2-TurboID sample (S9 Table). Among these, we focused on two putative RNA-binding proteins: the ZC3H3 ortholog Red5 (SJAG_00456) and the RBM26·27 ortholog Rmn1 (SJAG_03939). While orthologs of these proteins associate with Pab2 in S. pombe [16,31,38], they are dispensable for constitutive heterochromatin assembly in S. pombe [31]. We sought to determine their role(s) in S. japonicus.

Yeast two-hybrid assays (Y2H) confirmed that Pab2 interacts with both Red5 and Rmn1, and Red5 interacts with Rmn1 (Fig 6A and 6B). This suggests potential interactions that could form a trimeric complex. Growth on selective media (CM–Leu–Trp–His and CM–Leu–Trp–His–Ade) indicated that Pab2-Rmn1 and Red5-Rmn1 interactions are strong, whereas the Pab2 and Red5 interaction is relatively weak (Fig 6A and 6B). Consistent with these Y2H results, microscopic analyses showed that Pab2 colocalizes with both Red5 and Rmn1 (Fig 6C). Thus, we propose that Pab2 directly associates with Red5 and Rmn1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Two Pab2-binding partners, Red5 and Rmn1, are required for centromeric heterochromatin assembly.

(A and B) Yeast two-hybrid assays examining interactions between Pab2-Rmn1, Pab2-Red5, and Red5-Rmn1. Ten-fold serial dilutions of S. cerevisiae AH109 strains carrying indicated plasmids were spotted onto minimal medium plates (CM) lacking leucine and tryptophan (–Leu–Trp), leucine, tryptophan, and histidine (–Leu–Trp–His), or leucine, tryptophan, histidine, and adenine (–Leu–Trp–His–Ade). The tumor suppressor p53 and T antigen (T) served as positive controls (p53-T), while Lamin and T antigen served as negative controls (Lamin-T). DBD, DNA-binding domain; AD, activation domain. (C) Fluorescent microscopy images showing the colocalization of Pab2 with Red5 and Rmn1. The white dotted lines indicate cell shapes. Scale bars, 2 μm. (D–F) ChIP-qPCR analysis of H3K9me2 and total H3 enrichment at centromeres using the indicated strains. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (**p < 0.01, ***p < 0.001, and ****p < 0.0001; N.S.: not significant). (G) RT-qPCR analysis of transcripts derived from mei4⁺, ssm4⁺, spo5⁺, and rec8⁺, as well as centromeric (cen) and telomeric (tel) transcripts, in WT and pab2ΔN strains. The transcripts were normalized to act1 mRNA to determine their relative expression levels. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05, **p < 0.01, and ***p < 0.005; N.S.: not significant). (H) RNA-IP and ChIP-qPCR analysis of Pab2-GFP and Pab2∆N-GFP. (Left) RNA-IP followed by RT-qPCR to assess RNA-binding activity in cells expressing untagged Pab2 (No tag), Pab2-GFP, or Pab2∆N-GFP. Centromeric RNA (cen) was normalized to act1 mRNA to determine their relative enrichment. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test, with ‘Pab2’ as the reference sample (*p < 0.05). (Right) ChIP-qPCR analysis to assess Pab2-GFP enrichment at centromeres using the indicated strains. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with ‘No tag’ as the reference sample (*p < 0.05 and **p < 0.01; N.S.: not significant).

https://doi.org/10.1371/journal.pgen.1011647.g006

To determine the roles of Red5 and Rmn1 in constitutive heterochromatin formation in S. japonicus, we generated a red5–1 mutant and a rmn1Δ strain. Given that Red5 is essential in S. pombe, we isolated the red5–1 mutant in S. japonicus while avoiding potential lethality. The red5–1 mutant strain carries two amino acid substitutions (Gln253Arg and Thr265Ala) in the zinc-finger domain (S6A Fig), resulting in increased sensitivity to cold temperatures (S6B Fig). H3K9me2 ChIP-qPCR analysis revealed reduced levels of H3K9me2 at centromeres in red5–1 and rmn1Δ (Fig 6D and 6E). Moreover, there was no evident cumulative effect when we combined either red5–1 or rmn1Δ with pab2Δ (Fig 6D and 6E). Consistent with the decreased H3K9me2 levels, both red5–1 and rmn1Δ were sensitive to TBZ (S6C Fig) and exhibited increased levels of centromeric transcripts (S6D Fig). These results strongly suggest that Red5 and Rmn1 cooperate with Pab2 to assemble constitutive heterochromatin.

To further investigate the significance of Pab2’s interactions with Red5 and Rmn1 in heterochromatin assembly, we sought to determine the functional consequences of disrupting these interactions. Y2H assays using three Pab2 fragments (N-terminal domain, RRM, and IDR) revealed that the N-terminal domain of Pab2 is responsible for binding to Red5 and Rmn1 (S6E Fig), leading us to construct a strain that expresses Pab2 lacking the N-terminal domain (pab2∆N). Western blotting confirmed that Pab2ΔN-GFP expression levels were comparable to those of Pab2-GFP (S6F Fig). While Pab2ΔN-GFP failed to form nuclear foci, Red5 and Rmn1 still formed nuclear dots (S6G Fig), indicating that Red5 and Rmn1 localization does not depend on their interactions with Pab2. ChIP-qPCR analysis revealed reduced H3K9me2 levels at centromeres in pab2ΔN cells (Fig 6F), and pab2ΔN cells displayed elevated sensitivity to TBZ (S6H Fig), indicating a significant loss of heterochromatin. Furthermore, transcripts derived from mei4⁺, ssm4⁺, spo5⁺, and rec8⁺ were increased in pab2ΔN cells (Fig 6G). Intriguingly, Pab2∆N-GFP exhibited a substantial loss of RNA- and centromere-binding ability (Fig 6H). This suggests that the N-terminal domain of Pab2 plays a crucial role in its RNA and centromere binding. Further investigation is needed to determine whether this is due to direct binding, structural contributions, or modulation of interactions with Red5/Rmn1. Taken together, these findings demonstrate that the interaction between Pab2 and Red5/Rmn1 is essential for heterochromatin assembly.

The canonical poly(A) polymerase Pla1 is essential for the formation of constitutive heterochromatin

Studies in S. pombe have shown that centromeric transcripts are transcribed by RNA polymerase II (Pol II), suggesting that these transcripts are polyadenylated [43] and the Pol II C-terminal domain recruits the canonical poly(A) polymerase Pla1 to add poly(A) tail [71]. Given that poly(A) tails are generally recognized by poly(A)-binding proteins [72], we hypothesized that S. japonicus Pla1 (SJAG_02720) might indirectly contribute to heterochromatin assembly by facilitating Pab2 recruitment to these transcripts through polyadenylation. To test this, we analyzed a pla1 mutant strain (pla1–3), which contains multiple amino acid substitutions in both the N-terminal nucleotidyltransferase domain and the C-terminal RNA-binding domain (S10 Table). The pla1–3 mutant strain exhibited increased sensitivity to both low and high temperatures (Fig 7A). Consistent with a potential role for Pla1 in heterochromatin formation, ChIP-qPCR analysis indicated that H3K9me2 levels at centromeres were reduced in pla1–3 cells (Fig 7B). Additionally, pla1–3 cells showed greater sensitivity to TBZ than WT cells (Fig 7C). Moreover, RT-qPCR analysis demonstrated increased levels of transcripts derived from mei4⁺, spo5⁺, rec8⁺, and centromeres in pla1–3 (Fig 7D). Finally, yeast two-hybrid assays did not reveal a direct physical interaction between Pla1 and Pab2 (Fig 7E), suggesting that Pab2 and Pla1 likely function in distinct steps of the heterochromatin assembly pathway. These findings suggest a possible link between Pla1 function and constitutive heterochromatin formation in S. japonicus, potentially through its role in polyadenylation of centromeric transcripts and subsequent influence on Pab2 recruitment. Further investigation is required to fully elucidate the role of Pla1 in heterochromatin assembly.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Pla1 is required for centromeric heterochromatin.

(A) Growth assay of wild-type (WT), clr4–1 and pla1–3 cells. Ten-fold serial dilutions were spotted onto complete medium plates and incubated at the indicated temperatures. (B) ChIP-qPCR analysis of H3K9me2 and total H3 levels at centromeres in the indicated strains. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (***p < 0.001 and ****p < 0.0001; N.S.: not significant). (C) Thiabendazole (TBZ) sensitivity assay of WT, clr4–1, and pla1–3 cells. Ten-fold serial dilutions were spotted onto complete medium plates with or without TBZ and grown at 30°C. (D) RT-qPCR analysis of transcripts from mei4⁺, ssm4⁺, spo5⁺, and rec8⁺, as well as centromeric (cen) and telomeric (tel) transcripts, in WT and pla1–3. The transcripts were normalized to act1 mRNA to determine their relative expression levels. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a two-tailed unpaired t-test (*p < 0.05; N.S.: not significant). (E) Yeast two-hybrid assays examining the interaction between Pab2 and Pla1. Ten-fold serial dilutions of S. cerevisiae AH109 strains carrying indicated plasmids were spotted onto minimal medium plates (CM) lacking leucine and tryptophan (–Leu–Trp), leucine, tryptophan, and histidine (–Leu–Trp–His), or leucine, tryptophan, histidine, and adenine (–Leu–Trp–His–Ade). The tumor suppressor p53 and T antigen (T) served as positive controls (p53-T), while Lamin and T antigen served as negative controls (Lamin-T). DBD, DNA-binding domain; AD, activation domain.

https://doi.org/10.1371/journal.pgen.1011647.g007

Dri1 is dispensable for constitutive heterochromatin

Dri1 (SJAG_02381) was identified as a Pab2-proximal protein in our TurboID experiments (S9 Table). Dri1 is an RNA-binding protein with an RRM, and recent reports have demonstrated its involvement in heterochromatin assembly in S. pombe [73,74]. Yeast two-hybrid assays indicated slightly better growth of yeast cells carrying DBD-Rmn1 and AD-Dri1 on–Leu–Trp–His plates than those carrying DBD-p53 and AD-Dri1 (S7A Fig). This result suggests a possible, but weak, interaction between Rmn1 and Dri1. However, due to the qualitative nature of the assay, our observation does not provide sufficient evidence to confirm a direct and biologically relevant interaction between Rmn1 and Dri1. Further analyses are required to determine if there is any significant interaction. In addition, ChIP-qPCR analysis demonstrated that H3K9me2 levels at centromeres and telomeres in WT and dri1Δ were comparable (S7B Fig). Furthermore, dri1Δ cells were not sensitive to TBZ (S7C Fig). These results indicate that S. japonicus Dri1 is dispensable for centromeric and telomeric H3K9me2, highlighting differences in heterochromatin assembly mechanisms between S. japonicus and S. pombe.

Ppn1/PPP1R10, a CPSF constituent, interacts with Pab2 and the H3K9 methyltransferase Clr4/SUV39

Our TurboID approach also identified Ppn1 (SJAG_03500), the ortholog of PPP1R10 and a component of CPSF complex, as a Pab2-associated protein (S9 Table). In S. pombe, Ppn1 is part of the DPS (Dis2-Ppn1-Swd22) module within the CPSF complex [75]. While the involvement of CPSF in heterochromatin formation is established in S. pombe [26,27], the specific contribution of CPSF to this process remains unclear. To investigate the role of Ppn1 in S. japonicus heterochromatin assembly, we first examined its interaction with Pab2. Yeast two-hybrid assays confirmed the interaction between Pab2 and Ppn1 (Fig 8A). Notably, none of the Pab2 fragments (N, RRM, IDR, ΔN, ΔRRM, or ΔIDR) interacted with Ppn1 (S8A Fig), indicating that the full-length Pab2 is required for this interaction. Consistent with these results, fluorescent microscopy demonstrated that Ppn1 foci colocalize with Pab2 foci (Fig 8B). Furthermore, Ppn1 nuclear foci are largely absent in the absence of Pab2 (S8B Fig), while western blotting confirmed that Ppn1 expression levels in pab2Δ was comparable to those in WT (S8C Fig). These results further support a physical association between Pab2 and Ppn1. Remarkably, we found that Ppn1 and Rmn1 also directly interact with Clr4, as confirmed by yeast two-hybrid assays (Fig 8C). We also found that the Clr4 middle domain (amino acids: 76–220), but not the chromodomain (amino acids: 1–75) or SET domain (amino acids: 221–518), interacts with Ppn1 (S8D Fig). These findings suggest that Clr4, Pab2, and Ppn1 may function together within heterochromatic domains.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. The CPSF constituent Ppn1 is required for the assembly of constitutive heterochromatin.

(A) Yeast two-hybrid assays assessing the interaction between Pab2 and Ppn1. Ten-fold serial dilutions of S. cerevisiae AH109 strains carrying the indicated plasmids were spotted onto selective medium plates lacking leucine and tryptophan (–Leu–Trp), leucine, tryptophan, and histidine (–Leu–Trp–His), or leucine, tryptophan, histidine, and adenine (–Leu–Trp–His–Ade). p53-T antigen (T) and Lamin-T served as positive and negative controls, respectively. (B) Localization of Pab2-GFP and Ppn1-tdTomato was examined by fluorescent microscopy. The white dotted lines indicate cell shapes. Scale bars, 2 μm. (C) Yeast two-hybrid assays testing Ppn1-Clr4, Rmn1-Clr4, and Ppn1- Rmn1 interactions were performed as described in (A). (D) Schematic of Ppn1 showing the V354A and L355P substitutions, which disrupt the interaction with Clr4. Blue boxes indicate predicted intrinsically disordered regions. (E) Growth assay of wild-type (WT), clr4–1, and the ppn1^{V354, L355P} mutant cells. Ten-fold serial dilutions were spotted onto complete medium plates, and the plates were incubated at 30°C. (F) Western blotting of Ppn1- and Ppn1^{V354, L355P}-tdTomato, with Pcn1/PCNA as a loading control. (G) Localization of Ppn1- and Ppn1^{V354, L355P}-tdTomato in vegetative cells. The white dotted lines indicate cell shapes. Scale bars, 2 μm. (H) Levels of H3K9me2 and total H3 at centromeres and telomeres in WT, clr4–1, and ppn1^{V354, L355P} mutant strains assessed by ChIP-qPCR. Data are presented as mean ± SD from three independent experiments. Statistical significance was determined using a one-way ANOVA followed by Dunnett’s multiple comparison test, with WT as the reference sample (*p < 0.05, **p < 0.01, and ***p < 0.001; N.S.: not significant). (I) Thiabendazole (TBZ) sensitivity assay of WT, clr4–1, and ppn1^{V354, L355P} mutant cells. Ten-fold serial dilutions were spotted onto complete medium plates with or without TBZ and grown at 30°C.

https://doi.org/10.1371/journal.pgen.1011647.g008

Mutations in Ppn1 disrupt constitutive heterochromatin

To investigate Ppn1’s role in heterochromatin formation in S. japonicus, we initially attempted to generate a ppn1∆ strain. Although ppn1⁺ is non-essential for viability in S. pombe [76], we were unable to obtain a viable ppn1∆ strain in a wild-type S. japonicus strain. This observation, while requiring further investigation, hints at a potential difference in Ppn1 function or genetic context between the two species, possibly involving RNAi-dependent heterochromatin assembly as suggested by previous reports on RNAi components in S. japonicus [24].

Because we could not obtain a ppn1∆ strain, we next aimed to generate ppn1 mutant strains using PCR-based random mutagenesis [77]. Despite repeated attempts, we could not isolate any temperature- or cold-sensitive ppn1 mutants. Therefore, to specifically investigate the role of the Pab2-Ppn1 interaction in heterochromatin assembly, we employed a yeast reverse two-hybrid system [78] to select for ppn1 mutations that disrupt this interaction. This screen isolated a ppn1 mutant with two amino acid substitutions: Val354Ala and Leu355Pro (Fig 8D). According to AlphaFold predictions [79,80], these substitutions are located within an α-helix of Ppn1, and the proline residue (Leu355Pro) likely disrupts the helix structure [81]. Standard yeast two-hybrid assay showed that Ppn1^{V354A, L355P} lost its ability to interact with Clr4 and Pab2 (S8E Fig). Introduction of the V354A and L355P mutations into the endogenous ppn1⁺ gene resulted in a significant growth defect (Fig 8E). Western blotting revealed that the mutations reduced the steady-state level of Ppn1 (Fig 8F). Although Ppn1^{V354A, L355P}-tdTomato is detectable by western blotting, we did not observe any nuclear condensates (Fig 8G). This lack of localization, along with the reduced Ppn1 protein level (Fig 8F), suggests that the mutations disrupt Ppn1 localization and/or stability. In addition, ChIP-qPCR analysis showed decreased H3K9me2 levels at centromeres and telomeres in ppn1^{V354, L355P} mutant cells (Fig 8H). The ppn1^{V354, L355P} mutant also exhibited increased sensitivity to TBZ (Fig 8I), and transcripts derived from mei4⁺, rec8⁺, spo5⁺, and centromeres accumulated in the mutant cells (S8F Fig). These findings suggest that Ppn1 contributes to heterochromatin assembly, selective gene silencing, and maintenance of genome stability in S. japonicus.

Divergence between S. japonicus and S. pombe

The finding that Pab2 is essential for proper growth, efficient mating, and constitutive heterochromatin formation in S. japonicus represents a significant divergence from its role in S. pombe. While S. pombe Pab2 plays a role in suppressing meiotic gene expression [33,36], it does not seem essential for normal vegetative growth or efficient mating [33]. Furthermore, to the best of our knowledge, no prior studies have clearly demonstrated a role for Pab2 in the assembly of constitutive heterochromatin in S. pombe. This functional divergence suggests that Pab2 has acquired novel roles in S. japonicus. Several factors could contribute to these differences, including variations in genome organization [21] and/or differences in the mechanisms for cell growth and meiosis. These possible differences between the two species could also necessitate a more critical role for Pab2 in S. japonicus.

This divergence in Pab2 function is further exemplified by comparisons of heterochromatin assembly. In S. pombe, none of Red1, Red5, or Rmn1 is essential for H3K9 methylation at constitutive heterochromatic domains [31]. In contrast, Pab2, Red5, and Rmn1 are all required for centromeric heterochromatin in S. japonicus (Fig 3 and 6). Notably, we also found that Mit1 plays a critical role in H3K9 methylation at centromeres in S. japonicus (Fig 4), though not observe in S. pombe [15]. These differences in heterochromatin assembly mechanisms likely stem from variations in centromeric DNA sequences. S. japonicus centromeric DNA contains Ty3/Gypsy-type retrotransposons [21] and utilizes RNAi and heterochromatin for retrotransposon suppression [24]. Transposable elements are found at the centromeres of various eukaryotes [83–85], and RNAi acts as a defense system against them [86,87]. S. pombe, however, lacks such transposable elements within centromeres, instead using repetitive sequences called dg/dh [88]. This suggests that Pab2 in S. japonicus targets these centromeric retrotransposons. This idea can be supported by the involvement of Pab-1, a PABP in C. elegans, in transposon silencing [89] and the known connections between retrotransposons, the poly(A) tail, and PABPs [90–93]. It would be of interest to investigate the role of PABPs in heterochromatin assembly and transposon suppression in multicellular organisms.

Beyond centromeric structure, other differences exist between two species. Mitotic condensed chromosomes are visible in S. japonicus but not in S. pombe, and S. japonicus cells are larger [82]. These characteristics, along with the unique aspects of Pab2, highlight the value of S. japonicus as a model organism for uncovering novel aspects of conserved cellular processes not readily apparent in S. pombe. s its unique strength in studying cell biology and chromosome biology. While S. pombe research infrastructure is more developed, resources for S. japonicus are growing, including two programs we have developed [94], and we anticipate further development as the advantages of this organism are recognized.

Dissecting Pab2 functional domains

Pab2 comprises three domains: the N-terminal domain, the RNA recognition motif (RRM), and the C-terminal intrinsically disordered region (IDR). Yeast two-hybrid assays indicated that the N-terminal domain interacts with Red5 and Rmn1 (S6E Fig), suggesting its importance in forming the Pab2-Red5-Rmn1 complex. Consistent with this, deletion of the N-terminal domain (Pab2∆N) abolished nuclear dot formation and resulted in a substantial loss of centromeric RNA binding and reduced association with centromeres (Fig 6). These findings demonstrate that Red5 and Rmn1 are essential for the stable association of the Pab2 complex with centromeric RNA and heterochromatin, suggesting that the Pab2-Red5-Rmn1 complex, rather than Pab2 alone, is the primary functional unit at heterochromatic domains.

The RRM is a well-characterized RNA-binding domain [68], and we hypothesized that Pab2’s association with centromeric transcripts via its RRM is critical for constitutive heterochromatin formation. Mutational analysis of the RRM strongly supports this hypothesis. The Pab2^RRM3A mutant, which exhibited a substantial reduction in RNA-binding ability with detectable RNA binding, failed to form nuclear condensates and showed reduced levels of H3K9me2 at centromeres. This demonstrates that a critical threshold of RNA binding is necessary for Pab2 to form functional condensates and promote heterochromatin assembly. This heterochromatin defect, resulting from even partial disruption of RNA binding, highlights the importance of the Pab2-centromeric RNA interaction and suggests that centromeric RNAs may act as a structural component of the condensates, scaffolding the assembly of droplets containing heterochromatin factors as described previously [95,96].

IDRs are known to be crucial for biomolecular condensate formation through phase separation [54,55]. Our analyses demonstrated that the Arg residues within the IDR, which are crucial for the IDR interactions that assemble biological condensates [62], are essential for Pab2 droplet formation both in vitro and in vivo (Fig 4). RNA-IP data showed that Pab2^RRM3A still possess some RNA-binding ability, while Pab2^RRM3A-RA appears to have lost the ability completely. This indicates that the IDR also contributes independently to the association with centromeric RNAs, consistent with previous findings that Arg/Gly-rich (GAR) domains and RGG/RG boxes can bind nucleic acids [97]. Therefore, both RRM and IDR are involved in RNA association within Pab2 nuclear condensates.

Phase separation and heterochromatin assembly

The involvement of RNA-binding proteins and phase separation in heterochromatin assembly is increasingly recognized across diverse eukaryotes. In mammals, for example, RNA-binding proteins like FUS and hnRNP U have been implicated in heterochromatin formation [98,99], and phase separation of HP1 and other heterochromatin factors has been shown to be crucial for heterochromatin domain organization [61,100]. Our findings in S. japonicus further support these conserved principles, highlighting the importance of RNA and phase separation in heterochromatin regulation. However, while some aspects are conserved, the specific proteins involved and their precise roles can differ significantly between species. For example, while Pab2, Red5, and Rmn1 play a crucial role in S. japonicus heterochromatin assembly, their orthologs have a less critical role in S. pombe. This underscores the importance of studying diverse model organisms to fully understand the complexity of heterochromatin regulation.

Differential roles of Pab2 condensates

Our observation that both the RRM and IDR of Pab2 are required for the formation of nuclear condensates raises the question of how these structures contribute to Pab2’s diverse functions. Notably, the visible Pab2 condensates colocalize with Red1 (S1H Fig), a protein involved in selective mRNA decay but dispensable for heterochromatin assembly (Fig 3C). This suggests a role for these condensates in the mRNA degradation. Importantly, Pab2 condensates do not colocalize with Chp1 foci, which mark heterochromatic domains (S5C Fig). However, ChIP-qPCR data demonstrate that Pab2 also associates with centromeric regions (Fig 5F), suggesting a distinct interaction with heterochromatin, likely in a less concentrated manner. Given the essential role of centromeric RNA binding for heterochromatin assembly as demonstrated by our RRM mutant analysis (Fig 5), we propose that Pab2 can form distinct types of condensates with different properties: one type primarily involved in selective mRNA decay and another, dependent on centromeric RNAs, involved in heterochromatin assembly. Specifically, we hypothesize that the presence of centromeric RNAs alters the composition and/or properties of Pab2 condensates, enabling the recruitment or stabilization of key heterochromatin factors such as Clr4 and Mit1 at centromeric regions. This functional diversification of Pab2, mediated by distinct condensate properties, suggest a complex interplay between RNA metabolism and heterochromatin regulation.

Possible roles of Red5 and Rmn1 in heterochromatin assembly

While Red5 and Rmn1 associate with Pab2 in both S. japonicus (this study) and S. pombe [16,17,31], they are essential for proper heterochromatin assembly only in S. japonicus (Fig 6). However, their precise molecular functions in S. japonicus remain to be determined. It is possible that Red5 and Rmn1 are essential for stabilizing the association of the Pab2 complex with centromeric RNAs and heterochromatin, as discussed above. Furthermore, given that Pab2 exhibits preferential binding to putative meiotic mRNAs and centromeric RNAs (Fig 2D) despite its general affinity for poly(A) RNA, Red5 and Rmn1 could contribute to the RNA-binding specificity of the Pab2 complex. It is also plausible that Rmn1 (and potentially Red5) facilitates the recruitment or stabilization of other heterochromatin factors, such as Clr4 and Mit1, at heterochromatic domains. Finally, considering that Red5 and Rmn1 also contain intrinsically disordered regions and form nuclear foci independently of Pab2’s N-terminal domain (S6G Fig), they could contribute to the formation, stability, or organization of Pab2 condensates themselves. Future studies, such as co-immunoprecipitation experiments to examine the interactions of Red5/Rmn1 with other heterochromatin factors and RNA-binding assays using specific RNA sequences, will be crucial for elucidating the precise molecular functions of Red5 and Rmn1.

Ppn1 and its potential connection to heterochromatin assembly

Previous studies have implicated the CPSF complex in the formation of heterochromatin in S. pombe [26,27], but its role in heterochromatin assembly is less clear. Our finding that the CPSF component Ppn1 interacts with Clr4 raises the possibility that Ppn1 may also have a function at heterochromatic loci. Given the interaction of Ppn1 with Pab2 (Fig 8), we hypothesize that Ppn1 might modulate the function of Pab2 at these loci. For example, Ppn1 could influence the efficiency of Pab2 condensate formation or stability. Alternatively, Ppn1 could attract Clr4 and Mit1 to Pab2 condensates. However, our data only demonstrate interactions between Ppn1, Pab2, and Clr4, highlighting the need for further experiments to characterize Ppn1 and determine whether the entire CPSF complex is involved.

A model for Pab2-mediated heterochromatin assembly in S. japonicus

Based on our findings, we propose a working model for Pab2-mediated heterochromatin assembly at centromeres in S. japonicus (Fig 9). This model centers on the role of centromeric non-coding RNAs (ncRNAs) as a scaffold for heterochromatin formation. The Pab2 complex is recruited to heterochromatic domains via these ncRNAs. The RRM provides a crucial threshold of RNA binding, and the IDR further strengthens this interaction. Subsequent phase separation of Pab2, driven by IDR interactions, leads to the formation of nuclear condensates at heterochromatic domains. The bound ncRNAs likely become integral structural components of these condensates, contributing to their specific properties. This concept of ncRNAs acting as scaffolds for heterochromatin assembly is supported by a previous study in mammalian systems, where Suv39h (a mouse ortholog of Clr4) has been shown to directly bind ncRNAs derived from heterochromatic regions, leading to stabilized Suv39h localization at pericentromeric heterochromatin [101]. Pab2 condensates then promote heterochromatin assembly by concentrating heterochromatin factors such as Clr4 and Mit1 at these loci, thereby facilitating their activity. Moreover, Ppn1, by interacting with both Pab2 and Clr4, may coordinate their functions, potentially by stabilizing the interaction between Pab2 and Clr4. This working model highlights the interplay between ncRNA, phase separation, and protein-protein interactions in Pab2-mediated heterochromatin assembly.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 9. A model for Pab2-mediated heterochromatin assembly in S. japonicus.

https://doi.org/10.1371/journal.pgen.1011647.g009

The Pab2 complex (Pab2, Red5, and Rmn1) is recruited to heterochromatic regions via non-coding RNAs (ncRNAs) derived from these loci. The Pab2 complex then undergoes phase separation, forming nuclear condensates. These condensates concentrate key heterochromatin factors, such as the H3K9 methyltransferase Clr4 and the chromatin remodeler Mit1, at heterochromatic domains. This concentration facilitates their recruitment and/or stabilizes the association of these factors, thereby promoting heterochromatin assembly. In the absence of Pab2, heterochromatin assembly is significantly impaired.

Future directions

This study raises several important questions that warrant further investigation, both in S. japonicus and in other systems. First, directly testing the interaction between Pab2 and RNAi components in S. japonicus would be important to establish a concrete link between these pathways, given the established role of RNAi in heterochromatin formation in this organism. Second, investigating the in vivo dynamics of Pab2 condensates and their relationship to heterochromatin domains will further elucidate the mechanism of heterochromatin assembly. Third, determining the precise roles of Red5 and Rmn1 within the Pab2 complex, and exploring their potential interactions with other heterochromatin factors, such as those involved in histone modification, will also be valuable. Fourth, characterizing Mit1 function in constitutive heterochromatin assembly, which is not evident in S. pombe, will advance our understanding of the heterochromatin assembly pathway in S. japonicus. Finally, further studies are needed to investigate the role of polyadenylation, including the contribution of Pla1, in heterochromatin assembly and the recruitment of associated factors.

Given our findings demonstrate a role of Pab2 in heterochromatin formation in S. japonicus, it will be crucial to determine: (1) whether the functional interplay between PABP and heterochromatin assembly observed in S. japonicus is conserved in other species, (2) whether human PABPs similarly interact with factors involved in heterochromatin formation or the RNAi pathway, and (3) whether the principles governing the formation and function of Pab2 condensates found in S. japonicus are applicable to the organization of heterochromatin in more complex genomes. Addressing these questions will deepen our understanding of heterochromatin regulation and potentially reveal novel roles for PABPs beyond mRNA metabolism.

Media and S. japonicus strain construction

Strains were cultured in YES (Yeast Extract Supplement) rich medium. Gene deletion and C-terminal tagging strains of S. japonicus were generated using a PCR-based gene targeting method as described previously [102]. Primers for the PCR-based strain constructions were designed using Yesprit [94]. Specific point mutations (Pab2^RA, Pab2^RRM3A, Pab2^RRM3A-RA, and Ppn1^{V354A, L355P}) were introduced using the “pop-in, pop-out” allele replacement method [103]. Transformation of S. japonicus was performed as described previously [104]. DNA fragments used for transformation were amplified using PrimeSTAR Max DNA Polymerase (Takara Bio Inc.). Transformants were selected on YES plates containing an antibiotic or on Pombe minimal glutamate (PMG) medium plates lacking uracil. Correct transformants were confirmed by genomic PCR using 2 × Taq Master Mix (Novoprotein Scientific Inc.).

Temperature/cold-sensitive (ts/cs) mutant strains were generated using an error-prone PCR-based random mutagenesis method [77]. Briefly, a target gene (clr4⁺, red5⁺, or pla1⁺), including the drug resistance marker natMX6, was subjected to error-prone PCR using LA Taq DNA Polymerase (Takara Bio Inc.). To increase the mutation rate, dGTP was added to the PCR reaction to a final concentration of 0.8 mM. The resulting PCR products were used to transform a wild-type strain, and the transformants were selected on YES plates containing the appropriate antibiotics at 26°C for 3–4 days. Single colonies were replica-plated onto YES plates containing Phloxine B (PhB) and incubated at 18°C, 26°C, and 37°C. PhB stains dead cells deep red, whereas viable cells appear pink. Transformants exhibiting growth defects at 18°C and/or 37°C were subjected to ChIP-qPCR analysis of H3K9me2 to identify mutants defective in heterochromatin assembly (resulting in clr4–1, red5–1, and pla1–3). The mutations in the corresponding genes of three mutants were determined by Sanger sequencing of genomic DNA. S. japonicus strains used in this study are listed in S11 Table.

Western blotting

Protein extracts were prepared using previously described methods [39,105]. For detection of epitope tagged proteins, anti-GFP (No.11814460001, Roche), anti-c-myc (HT101–01, Transgene), anti-HA (HT301–01, Transgene), or anti-mCherry (bsm-33131M, Bioss) antibody was used. The S. japonicus PCNA ortholog Pcn1 was detected using an anti-PCNA antibody (A13336, ABclonal), which served as a protein loading control.

Fluorescent microscopy

Fluorescence and differential interference contrast (DIC) images were acquired using a Zeiss Axio Imager Z2 Upright Microscope (Carl Zeiss MicroImaging). ZEN lite 2012 software (Carl Zeiss MicroImaging) was used to process RAW images.

RNA analyses

Yeast cells were harvested at mid log-phase (OD₆₀₀ = 0.5–1). Total RNA was isolated using the MasterPure Yeast RNA Purification Kit (Lucigen). Strand-specific RNA-Seq was conducted using the VAHTS Universal V8 RNA-seq Library Prep Kit for Illumina (NR605–02, Vazyme) and NovaSeq 6000 S4 Reagent Kit (Illumina). Raw sequencing reads were processed as described previously [37]. The RNA-seq datasets generated in this study have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE237964. Gene ontology (GO) enrichment analysis was performed using Metascape [106]. Due to the limited availability of S. japonicus gene annotations in Matascape, orthologous genes in S. pombe were used for this analysis. RT-qPCR was performed using the PrimeScript II 1st strand cDNA Synthesis Kit (Takara Bio Inc.) and ChamQ Universal SYBR qPCR Master Mix (Vazyme). Primers used for qPCR are listed in S12 Table.

Chromatin Immunoprecipitation (ChIP)

ChIP was performed as described previously [37,107] with slight modifications. Specifically, fission yeast cells were grown to mid log-phase (OD₆₀₀ = 0.5–1) at 26°C (permissive temperature for all mutant strains) and harvested. Cells were washed with 1 × PBS and then fixed in 3% formaldehyde for 10 minutes at room temperature. Chromatin was sheared by sonication using a Bioruptor Plus sonication device (Diagenode) to an average size of less than 500 base pairs. For immunoprecipitation, 2 μg of antibodies against H3K9me2 (ab1220, Abcam), H3K9me3 (DF6938, Affinity Biosciences), histone H3 (ab1791, Abcam), or anti-GFP (ab290, Abcam) antibody as well as 20 μL of protein A/G magnetic beads (P2108, Beyotime Biotechnology) were used for each sample. After immunoprecipitation, samples were reverse crosslinked by incubation at 68°C overnight. Primer sequences used for ChIP-qPCR are listed in S12 Table.

Identification of Pab2-binding proteins by TurboID and mass spectrometry

Proximity-dependent biotin labeling by TurboID in the search for Pab2-binding proteins was carried out as described previously [70]. Specifically, an S. japonicus strain expressing Pab2-TurboID and the corresponding parental strains were lysed using RIPA buffer [50 mM Tris−HCl (pH 7.5), 150 mM NaCl, 1.5 mM MgCl₂, 1 mM EGTA, 0.1% SDS, and 1% NP-40] supplemented with 0.4% sodium deoxycholate, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride (PMSF). Biotinylated proteins in the clarified lysates were purified using streptavidin magnetic beads (GE Healthcare Life Science), digested with trypsin (Promega), and desalted using SOLAµ SPE Plates (Thermo Fisher Scientific). Liquid chromatography with tandem mass spectrometry (LC-MS/MS) was performed using a Q Exactive HF-X mass spectrometer coupled to an Easy nLC 1200 system (Thermo Fisher Scientific). Peptides were separated using a 60-minute gradient on a homemade C18 column and ionized by electrospray. Full MS and HCD MS/MS spectra were acquired. Raw data were processed using MaxQuant 1.6.5.0, searching against the S. japonicus UniProt proteome (Proteome ID: UP000001744). Peptide and protein identifications were filtered to a 1% false discovery rate (FDR). Detailed parameters for LC-MS/MS and proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD053432.

Yeast two-hybrid assays

Yeast two-hybrid assays were performed using the pGBKT7 and pGADT7 plasmids that contain full-length coding sequences of the proteins of interest or those with specific mutations. The indicated plasmid pairs were co-transformed into the S. cerevisiae AH109 strain (MATa, trp1–901, leu2–3, 112, ura3–52, his3–200, gal4∆, gal80∆, LYS2::GAL1_UAS-GAL1_TATA-HIS3, MEL1, GAL2_UAS-GAL2_TATA-ADE2, URA3:: MEL1_UAS-MEL1_TATA-lacZ) (YC1010, Weidi Biotechnology Company). Successful co-transformation was confirmed by growth on complete minimal (CM) medium plates [108] lacking leucine and tryptophan (–Leu–Trp). Interactions were assessed by growth on CM medium lacking leucine, tryptophan, and histidine (–Leu–Trp–His) for medium interaction and on CM medium lacking leucine, tryptophan, histidine, and adenine (–Leu–Trp–His–Ade) for strong interactions.

A yeast reverse two-hybrid assay [78] was performed to identify critical amino acid residues of Ppn1 required for interaction with Clr4. Briefly, a cDNA fragment encoding Ppn1 was amplified by error-prone PCR using LA Taq DNA Polymerase (Takara Bio Inc.). The mutagenized ppn1⁺ fragment, a gapped pGBKT7 plasmid, and pGADT7 carrying the clr4⁺ gene were co-transformed into AH109. Transformants grown on CM–Leu–Trp plates were screened on CM–Leu–Trp–His plates for loss of growth, indicating loss of interaction. The ppn1 mutations from clones exhibiting compromised Ppn1-Clr4 interaction were determined by Sanger sequencing.

RNA immunoprecipitation (RNA-IP)

RNA-IP was conducted with anti-GFP (ab290, Abcam) antibody as described previously [39]. Specifically, yeast cells harvested during log-phase growth (OD₆₀₀ = 0.5–1) were washed with 1 × PBS containing 0.1 mM PMSF. Cells were lysed in RNA-IP buffer (50 mM HEPES, 140 mM NaCl, 10% glycerol, 1 mM EDTA, 0.1% Triton X-100, 0.1% NP- 40, 1 mM PMSF, 2 mM vanadyl ribonucleoside complex, 400 U/mL RNasin Plus RNase inhibitor) and 5% of the supernatant was saved as input after having been centrifuged. The remainder of the supernatant was then incubated with 40 μL of protein A/G magnetic beads (B23201, Bimake.com) coupled with 2 μL of anti-GFP antibody (ab290, Abcam) for 3 hours at 4°C with rotation. The beads were then separated and washed three times with RNA-IP buffer. RNA was cleaned up using the Trizol method, and DNA was removed through digestion with TURBO DNase (Thermo Fisher Scientific). Purified RNAs were again cleaned up using MasterPure Yeast RNA Purification Kit (Lucigen). RT-qPCR was performed using the PrimeScript II 1st strand cDNA Synthesis Kit (TaKaRa Bio Inc.) and ChamQ Universal SYBR qPCR Master Mix (Vazyme). Primer sequences used in RNA-IP are listed in S12 Table.

Recombinant protein expression and purification

MBP-Pab2–6 × His tag and MBP-Pab2RA-6 × His tag fusion proteins were expressed and purified from E. coli BL21-CodonPlus (DE3)-RIL cells (EC1008, Weidi Biotechnology Company). E. coli cells were pre-cultured overnight at 37°C and inoculated at a ratio of 5–10% into 500ml of LB medium containing the appropriate antibiotics. E. coli cells were grown to OD₆₀₀ = 0.5−0.8 at 37°C, 150 rpm and induced with 1 mM IPTG at 37°C, 120 rpm for 2–4 hours. Pelleted cells were lysed in lysis buffer (50 mM pH 7.5 Tris–HCl, 1M NaCl, 1 M urea, 10 mM imidazole, 1.5 mM 2-mercaptoethanol, 1% NP-40, 5% glycerol and protease inhibitor cocktail) by sonication. Lysates were pelleted at 15,000 rpm at 4°C for 45 min. Supernatants were purified using 2 mL Ni NTA Beads (SA004025, Smart-Lifesciences) prewashed with lysis buffer at room temperature. Supernatants incubated with beads at 4°C for 1 hour and then beads were washed with lysis buffer for 6 times. Proteins were eluted with elution buffer (50 mM pH 7.5 Tris–HCl, 1M NaCl, 1 M urea, 500 mM imidazole, 1.5 mM 2-mercaptoethanol and 5% glycerol). The eluted proteins were analyzed by SDS-PAGE and dialyzed into dialysis buffer (20 mM pH 7.5 Tris–HCl, 150 mM NaCl, and 1 mM DTT). Proteins in dialysis buffer were concentrated by Amicon Ultra-4 Centrifugal Filter Unit (UFC803008, Sigma) and then incubated with TEV protease at 30°C for 1 hour. The fractions were analyzed by SDS-PAGE and stored at −80°C after flash frozen in liquid nitrogen.

In vitro liquid droplet formation assay

After removing MBP and His tag in MBP-Pab2–6 × His tag and MBP-Pab2^RA-6 × His tag fusion proteins by TEV protease, proteins were diluted to the indicated concentrations with dialysis buffer (20 mM pH 7.5 Tris–HCl, 150 mM NaCl, and 1 mM DTT). Proteins were observed under a DIC microscope using a Zeiss Axio Imager Z2 microscope (Carl Zeiss MicroImaging) with a 40x objective. All imaged were captured within 5 min after LLPS induction at room temperature.