Sgf73, a subunit of SAGA complex, is required for the assembly of RITS complex in fission yeast

RNA interference (RNAi) is a widespread gene-silencing mechanism and is required for heterochromatin assembly in a variety of organisms. The RNA-induced transcriptional silencing complex (RITS), composed of Ago1, Tas3 and Chp1, is a key component of RNAi machinery in fission yeast that connects short interference RNA (siRNA) and heterochromatin formation. However, the process by which RITS is assembled is not well understood. Here, we identified Sgf73, a subunit of the SAGA co-transcriptional complex, is required for pericentromeric heterochromatin silencing and the generation of siRNA. This novel role of Sgf73 is independent of enzymatic activities or structural integrity of SAGA. Instead, Sgf73 is physically associated with Ago1 and Chp1. The interactions among the subunits of the RITS, including those between Tas3 and Chp1, between Chp1 and Ago1, between Ago1 and Tas3, were all impaired by the deletion of sgf73+. Consistently, the recruitment of Ago1 and Chp1 to the pericentromeric region was abolished in sgf73Δ cells. Our study unveils a moonlighting function of a SAGA subunit. It suggests Sgf73 is a novel factor that promotes assembly of RITS and RNAi-mediated heterochromatin formation.


Results and Discussion
Sgf73 is required for heterochromatin silencing. Our previous study showed that sgf73Δ cells are sensitive to thiabendazole (TBZ), a microtubule destabilizing drug 26 . Hypersensitivity to TBZ could arise from defects in centromeric heterochromatin, because heterochromatin is required to attract cohesin and ensure proper chromosome segregation 27 . To validate the potential role of Sgf73 in heterochromatin organization, we used a strain in which a ura4 + marker gene was inserted into the outermost (otr) pericentromeric heterochromatin of chromosome 1 (otr1R::ura4 + ) (Fig. 1a) 28 . Just as with cells lacking an essential effector of the RNAi machinery (dcr1Δ ), deletion of sgf73 + (sgf73Δ ) derepressed the ura4 + gene, resulting in poor growth on medium containing counterselective drug 5-fluoroorotic acid (5FOA) Figure 1. Sgf73 is required for heterochromatin silencing at pericentromeric region. (a) A schematic representation of centromere 1 and the position of a inserted marker gene (otr1R::ura4 + ). (b) Fivefold serial dilution assay to examine the silencing of otr1R::ura4 + . Wild-type (WT) cells with silenced ura4 + grow normally on medium containing 5FOA, while loss of silencing kills cells on 5FOA. (c) RT-PCR analysis of otr1R::ura4 + and pericentromeric repeat (dh) RNA levels relative to a control act1 + . The relative level in WT cells was arbitrarily designated as 1. Each column shown in (c) and below represents the mean ± s.d. from three biological repeats. (d,e) ChIP analysis of enrichment of Pol II (d) and H3K9me2 (e) at otr1R::ura4 + and dh relative to fbp1 + . Relative enrichment in WT cells was arbitrarily designated as 1.
Scientific RepoRts | 5:14707 | DOi: 10.1038/srep14707 and good growth on medium without uracil (Fig. 1b). Consistently, the transcripts from the otr1R::ura4 + and endogenous pericentromeric repeat (dh), and the occupancy of RNA polymerase II at both loci increased substantially in sgf73Δ cells, indicating impaired heterochromatin silencing at the pericentromeric region (Fig. 1c,d). A hallmark of heterochromatin is the presence of histone H3K9 methylation 29 . Like dcr1Δ cells, sgf73Δ cells exhibited reduced levels of H3K9 dimethylation (H3K9me2) at centromeric repeats (dh) and at the inserted marker ura4 + (Fig. 1e). Thus, we conclude that sgf73 + is essential for maintaining heterochromatin silencing and repressive histone modifications at the pericentromeric region.
Sgf73 mediates heterochromatin silencing in a SAGA-independent manner. Sgf73, along with the subunits of the HAT module of the SAGA complex, including Gcn5, Ada2, Ada3 and Sgf29, were shown to act as anti-silencing factors to prevent the spreading of heterochromatin in budding yeast 25,[30][31][32][33] . Unexpectedly, Sgf73 was found to promote silencing in fission yeast in this study (Fig. 1). Therefore, it was of interest to investigate whether the role of Sgf73 in heterochromatin silencing was mediated in the context of the SAGA complex.
SAGA is responsible for the transcription of a broad range of genes. For example, 10% of genes in budding yeast are subjected to regulation by SAGA upon stress response 34 . However, the mRNA levels of several factors critical for the heterochromatin assembly in sgf73Δ cells were not substantially different with those in WT cells (Fig. 2a, Supplementary Fig. S1). The factors subjected to assay include Ago1, Dcr1, Clr4, Rik1, Swi6, Clr3, Arb1, Arb2, Chp1, Cid12, Raf1, Stc1, Hrr1, Rdp1, Tas3, Clr1, Swi6 and Dsh1. These data are consistent with a previous microarray analysis 35 . They suggest that Sgf73 regulates heterochromatin silencing without affecting the transcriptions of these essential factors.
To identify the contributions of SAGA to heterochromatin silencing, representative subunits of SAGA were deleted in a strain carrying an otr1R::ura4 + reporter system. In contrast to a severe silencing defect observed in sgf73Δ cells, deletion of other subunit of the UBP module, including Sus1, Sgf11 or the The relative level to a control act1 + in WT cells was arbitrarily designated as 1. Each column shown in (a) and below represents the mean ± s.d. from three biological repeats. (b) Fivefold serial dilution assay to examine the silencing of otr1R::ura4 + in deletion mutants of representative SAGA subunits. (c) RT-PCR analysis of otr1R::ura4 + and pericentromeric repeat (dh) RNA levels relative to a control act1 + . (d) ChIP analysis of enrichment of Spt7-3HA at dh, dg, otr1R::ura4 + and mae2 + relative to fbp1 + . Relative enrichment in the cells without tagging (no tag) was arbitrarily designated as 1.
catalytic subunit Ubp8, did not affect silencing of ura4 + (Fig. 2b). Similarly, deletion of subunits of the HAT module, including Sgf29 or the catalytic subunit Gnc5, exhibited no silencing defect (Fig. 2b). Spt7 is a pivotal structural subunit of the SAGA complex. Deletion of Spt7 causes the dissociation of most modules and thus abolishes the activity of the SAGA complex 36 . As shown in Fig. 2b, albeit suffering a severe growth defect, spt7Δ cells maintained normal silencing of ura4 + . Consistently, transcript levels of dh and otr1R::ura4 + in the selected SAGA subunit mutants were kept low as in WT cells, except in the sgf73Δ cells (Fig. 2c). Therefore, these results suggest enzymatic activities and structural integrity of SAGA are not required for the heterochromatin silencing at pericentromeric region.
In a ChIP assay, Spt7 with a C-terminal triple HA tag (Spt7-3HA) was enriched at mae2 + , a gene targeted by the SAGA complex 35 , but not at the centromeric repeat (dh, dg) or inserted ura4 + (Fig. 2d). This result suggests that Spt7, probably in the context of SAGA, is not localized to the pericentromeric heterochromatin region. This is contrast with the enrichment of Sgf73 in heterochromatin shown in Fig. 3 below. Different behaviors between Sgf73 and other subunits of SAGA suggest the role of Sgf73 in the heterochromatin silencing is independent of SAGA. This novel role of Sgf73 might have evolved in fission yeast to cope with different heterochromatin effectors not present in Saccharomyces cerevisiae, such as RNAi 37 . On the other hand, a SAGA-independent role of Sgf73 is not an exceptional case in fission yeast. Ataxin-7, a human homologue of yeast Sgf73, is associated with microtubules and enhances the stability of microtubule filaments 38 . Sgf73 is an essential component of RNAi machinery. Formation and maintenance of centromeric heterochromatin requires RNAi 3 . As shown in Fig. 1, defective heterochromatin observed in the sgf73Δ cells was similar to that in dcr1Δ cells, suggesting Sgf73 might be involved in the RNAi pathway. Consistent with this idea, siRNAs corresponding to the pericentromeric repeats (dg, dh) were substantially decreased in sgf73Δ cells, albeit not totally abolished as in dcr1Δ cells, while the level of non-coding snoRNA U24 was not affected (Fig. 3a, Supplementary Fig. S2). This suggests Sgf73 is important, but not indispensable for the production of centromeric siRNA.
The involvement of Sgf73 in RNAi machinery was further investigated over the mating type region (mat2P/mat3M), where RNAi and Atf1/Pcr1 act in parallel pathways to nucleate heterochromatin formation 39 . Deletion of a key component of either pathway, such as Pcr1 or Ago1, did not affect the silencing of a marker ura4 + gene inserted into the K region (kint2::ura4 + ) (Fig. 3b,c). But disruption of both pathways, as demonstrated by a pcr1Δ ago1Δ mutant, resulted in the derepression of ura4 + marker and a severe growth defect on 5FOA (Fig. 3c). sgf73Δ and ago1Δ sgf73Δ cells exhibited normal silencing of ura4 + , but pcr1Δ sgf73Δ cells were killed on 5FOA due to the derepression of ura4 + (Fig. 3c). Accordingly, the RNA level of ura4 + in pcr1Δ sgf73Δ cells was substantially elevated, as observed in pcr1Δ ago1Δ cells (Fig. 3d). Overlapping of phenotype of sgf73Δ and ago1Δ cells in heterochromatin silencing over the mating type region strongly suggests Sgf73 acts in the RNAi pathway.
The relationships between Sgf73 and other RNAi components were validated by a complementation assay. The silencing defect of otr1R::ura4 + in sgf73Δ cells was substantially suppressed by overexpressing of dcr1 + , ago1 + or clr4 + , as shown by the improved growth on 5FOA (Fig. 3e). Accordingly, transcript levels of ura4 + and centromeric repeats (dh, dg) decreased in sgf73Δ cells overexpressing these factors (Fig. 3f). The result suggests Sgf73 is functionally linked with these key players during RNAi-mediated silencing.

Sgf73 is required for the assembly and recruitment of RITS complex. The localization of Sgf73
at the heterochromatin region suggests Sgf73 is physically associated with other RNAi components. As shown in Fig. 4a, Ago1 and Chp1, subunits of RITS complex, were co-purified with Sgf73-3HA in a co-immunoprecipitation assay. This suggests that Sgf73 is associated with RITS. Since Sgf73 was not among the peptides identified by tandem affinity purification using Chp1 as bait 15 , Sgf73 is unlikely to be a fundamental subunit of the RITS complex. Instead, Sgf73 is perhaps a peripheral component of RITS. Notably, the interactions among the subunits of the RITS, including that between Tas3 and Chp1, between Chp1 and Ago1, were severely impaired by the deletion of sgf73 + (Fig. 4b). Ago1-Tas3 interaction was also disrupted, but to a much lesser extent, in sgf73Δ cells (Fig. 4b). Sgf73 is the first known factor that regulates the integrity of RITS. In contrast, integrity of RITS is not dependent on the production of siRNA or H3K9 methylation, as the Tas3-Ago1-Chp1 composition of RITS is not affected by the deletion of dcr1 + or clr4 + 15 . A previous study indicates that the physical interaction between Ago1 and Tas3 is required for the recruitment of Ago1 to centromeres for efficient de novo establishment of centromeric heterochromatin 41 . Accordingly, the enrichments of Ago1 at the centromeric repeat (dh) and inserted ura4 + marker decreased substantially in sgf73Δ cells (Fig. 4c). The enrichment of Chp1 at both Scientific RepoRts | 5:14707 | DOi: 10.1038/srep14707 loci also decreased, but to a lesser extent comparing to those of Ago1, which might be explained by the tethering of Chp1 with residual H3K9me2 in sgf73Δ cells 17 . The results suggest that Sgf73 is critical for the recruitment of RITS to the centromeric heterochromatin.

Conclusions
In this study, we revealed an essential role of Sgf73 in RNAi-mediated heterochromatin silencing. Sgf73 is physically associated with RITS and is required for the integrity of RITS. Given the fact that the localizations of Ago1 and Chp1 at heterochromatin were both impaired in sgf73Δ cells, it suggests that RITS need to be fully assembled before it is recruited to the centromeric region to initiate heterochromatin (c) Five-fold serial dilution assay to examine the silencing of kint2::ura4 + in sgf73Δ cells, and in strains combining deletion of Sgf73 and deletion of a effector in the RNAi or Atf1/Pcr1 pathway. (d) RT-PCR analysis of kint2::ura4 + RNA levels from the strains in (c). The relative level to a control act1 + in WT cells was arbitrarily designated as 1 Each column shown in (d) and below represents the mean ± s.d. from three biological repeats. (e) Fivefold serial dilution assay to examine the silencing of otr1R::ura4 + in sgf73Δ cells overexpressing dcr1 + , ago1 + or clr4 + . sgf73Δ cells were transformed with a plasmid overexpressing the indicated gene and transformants were subject to the silencing assay. (f) RT-PCR analysis of dh, dg and otr1R::ura4 + RNA levels relative to a control act1 + in the transformants in (e). (g) ChIP analysis of enrichment of Sgf73-3HA at dh and otr1R::ura4 + relative to fbp1 + . Relative enrichment in the cells without tagging (no tag) was arbitrarily designated as 1.
silencing. However, the structural integrity of RITS is not required afterwards to maintain the localization of individual subunits at the centromeric region, as demonstrated by the phenotype of a Tas3 mutant which failed to interact with Ago1 41 . The recruitment of Sgf73 to the centromeric region is probably mediated through the associations with RITS, as the localization of Sgf73 relies on Dcr1 that produces siRNAs to guide RITS.
This study also extends a growing list of the moonlighting functions of SAGA subunits. Besides Sgf73, Spt20, a structural subunit of SAGA, regulates septin ring assembly through physical interactions with septins 42 . p38IP, a human homologue of Spt20, interacts with mammalian (m)Atg9 (ATG9A) and inhibits the trafficking of mAtg9 43 . Tra1, the largest subunit of SAGA, was identified as a subunit of a novel complex called ASTRA in fission yeast 44 . Divergently evolved functions of SAGA subunits reflect the capacity of a subunit in a large complex to acquire additional functions to adapt to new environments.

Methods
Yeast strains and plasmids. All the strains used in this study are listed in Supplementary Table   S1. Gene deletion and tagging were performed by homologous recombination using a plasmid-based method 45 . Cells were grown in yeast extract medium with supplements (YES) or Edinburgh minimal glutamate medium minus leucine (EMMG-Leu) medium 46 . ORF of sgf73 + , ago1 + , dcr1 + or clr4 + was cloned into pRep41 vector for overexpressing 47 . Fivefold serial dilution assay. Exponentially growing cells were collected and adjusted to an A 600 of 1.0. Samples were diluted by fivefold for five times. 5 μ l dilutions were spotted onto YES or EMMG-Leu medium supplemented with 5FOA (YY12210, Yuanye Biotechnology, Shanghai, China, 1 g/Liter) as indicated. Plates were incubated for 2 or 3d at 32 °C before imaging.  (a) Co-immunoprecipitation (Co-IP) assay to analyze the physical association between Sgf73 and subunits of RITS. Sgf73-3HA immunoprecipitation (IP) was followed by the Western blot (WB) of Ago1. Chp1-3FLAG IP was followed by the WB of Sgf73-3HA. Input and IP samples were run in the same gel. Cropped blots were shown for clarity. Full-length blots are presented in Supplementary Figs S4 and S5. (b) Co-IP assay to analyze the physical association between subunits of RITS in WT and sgf73Δ cells. Tas3-3FLAG IP was followed by the WB of Ago1 and Chp1. Chp1-3FLAG IP was followed by the WB of Ago1. Input and IP samples were run in the same gel. Cropped blots were shown for clarity. Full-length blots are presented in Supplementary Figs S6 and S7. (c) ChIP analysis of enrichment of Ago1 or Chp1 at dh and otr1R::ura4 + relative to fbp1 + . Relative enrichment in the WT cells was arbitrarily designated as 1 Each column represents the mean ± s.d. (n = 3) from three biological repeats. Co-immunoprecipitation. ~5 × 10 10 exponentially growing cells were harvested. Cells were washed and then resuspended in Buffer 1 supplemented with protease inhibitors as described above. Cells were broken by a high-pressure cell homogenizer (JN-02C, JNBIO, Guangzhou, China) and supernatant were collected after centrifugation. Supernatant was incubated with anti-FLAG M2 Magnetic beads (M8823, Sigma-Aldrich), or with anti-HA antibody and then protein A/G PLUS-agarose beads (sc-2003, Santa Cruz, Dallas, TX, USA) for 6h. Bead-conjugated complexes were washed with Buffer 1, boiled in SDS-gel loading buffer and subjected to Western blot.

RT
Northern blot. Small RNA fractions were prepared from exponentially growing cells using mirVana miRNA Isolation kit (AM1560, Life Technologies). Small RNA was resuspended in 50% formamide and separated on a 15% urea-denaturing poly-acrylamide gel. Samples on the gel were blotted onto a Hybond-N membrane (RPN303N, GE Healthcare, Piscataway, NJ, USA) using a semi-dry electrophoretic transfer cell (170-3912, Bio-Rad, California, USA). 32 P labeled oligonucleotide probes complimentary to dh and dg were generated by PCR. Probe against snoU24 was generated by end-labelling. Probes were hybridized to the membrane overnight at 38 °C in a rotating oven. The membrane was then washed twice and exposed to an X-ray film for 3 d at − 80 °C. Primers and oligos used are listed in Supplementary  Table S2.