| Edited by Hiroshi Iwasaki. Masaya Oki: Corresponding author. E-mail: ma4sa6ya@u-fukui.ac.jp |
Eukaryotic chromosomes are comprised of accessible euchromatic domains contained in loose chromatin structures and inaccessible heterochromatic domains within condensed structures that restrict the expression of internal genes. In S. cerevisiae, heterochromatin-like regions have been found at HM silent mating-type loci (HMR and HML), the telomeres, and in rRNA-encoding DNA. This transcriptional silencing is achieved and spread through propagation of the SIR (silent information regulator) protein complex, composed of Sir2p, Sir3p, and Sir4p. In the Sir complex, Sir2p has histone deacetylation activity (Imai et al., 2000) and moves along the DNA by deacetylating the N-terminal of histone to allow the other Sir proteins to bind to the nucleosomes (Rusche et al., 2003; Sun et al., 2011). While Sir2p, Sir3p, and Sir4p are all required for silencing at the HMR region, HML region and the telomeres, silencing of genes at the rDNA region requires only Sir2p (Rusche et al., 2003).
To regulate transcriptional silencing, boundaries are necessary to prevent uncontrolled gene silencing. Thus far, two models have been proposed to explain how silencing might be limited; DNA sequence-dependent and DNA sequence-independent. In the former model, STAR (subtelomeric anti-silencing regions) sequences near the telomeres (Fourel et al., 1999) and tDNA sequence at the right side of the HMR domain regulate the spread of silencing (Dhillon et al., 2009; Donze et al., 1999; Donze and Kamakaka, 2001, 2002; Oki and Kamakaka, 2005). The DNA sequence-independent model, on the other hand, relies on competition between histone deacetylation by Sir2p and histone acetylation by Sas2p, a subunit of the SAS-I complex, to define the silencing boundaries (Kimura and Horikoshi, 2004; Kimura et al., 2002; Suka et al., 2002).
We developed a genome-wide screening system to investigate whether the Gal4p DNA-binding domain (GBD) fused protein maintains boundary function during the silencing process. We isolated 55 boundary related genes from approximately 6,000 genes in S. cerevisiae (Oki et al., 2004). The function of one of these boundary proteins (Fub1p/Ycr076cp) was still unknown. In the past syudies, Fub1p had been shown to associate to more than 20 proteins, for example, proteasome associated proteins, histone modification proteins, transcriptional regulator proteins, small G-protein regulator proteins, cell cycle check point proteins, ER and/or Golgi related proteins, mitochondria related proteins and transport factor proteins with many kinds of comprehensive screening (Collins et al., 2007; Costanzo et al., 2010; Krogan et al., 2006). Interestingly, Fub1p has a PI31 proteasome regulator domain with higher eukaryote protein, and it is known to inhibit proteasome activity (Kirk et al., 2008; Mccutchen-Maloney et al., 2000; Zaiss et al., 1999, 2002). The proteasome is a proteolytic enzyme comprised of multiple component particles that degrades unneeded or damaged proteins with a polyubiquitinated targeting sequence (Hochstrasser, 1996). The 26S proteasome is comprised of more than 33 different subunits and is the most common form of proteasome. It contains one 20S proteasome core particle (20S CP) and two 19S regulatory particles (19S RPs) with multiple ubiquitin-binding sites that allow polyubiquitinated proteins to enter into the catalytic core (Baumeister et al., 1998; Coux et al., 1996; Voges et al., 1999). The 20S CP consists of four seven-fold symmetric rings with two outer α rings and two inner β rings (Groll et al., 1997). The 19S RP consists of two subassemblies: a lid that associates deubiquitination, and a base that binds to the 20S CP directly and maintains six ATPases that support the rapid degradation of unfolded proteins (Glickman et al., 1998). However, new attention is being paid to the non-proteolytic functions of the proteasome that have recently been reported, such as regulation of transcription (Szutorisz et al., 2006).
Thus far, a relationship between the proteasome and transcriptional silencing has not been reported. This study focuses on analyzing the relationship between Fub1p, which plays a role in the establishment of transcriptional silencing boundaries, and the proteasome complex, which interacts directly with Fub1p.
The yeast strains and plasmids used in this experiment are listed in Table 1 and Table 2. All domain analysis plasmids were generated by PCR-mediated sequence deletion to obtain plasmids in which specific amino acids were deleted. Using a pair of phosphorylated primers designed against sequences just outside the target domain, the remainder of the plasmid was amplified and ligated, effectively deleting the target amino acids.
 View Details | Table 1 Yeast strains |
 View Details | Table 2 Plasmid |
A human cDNA library was prepared by RT-PCR using mRNA from mesenchymal stem cells (MSCs) as a template. An EcoRI/NotI PSMF1 fragment was amplified by PCR using the primers 5’- aattagaattcatggcgggcctgga - 3’ and 5’- ttaaagcggccgctcacaggtaca - 3’ (underlined areas indicate restriction sites) and cloned into a pGBK-RC plasmid (Ito et al., 2000) to generate pFOM279.
To construct the gene disruption strains, the genes were replaced by the KanMX gene (Yeast Knock Out Strain Collection, Open Biosystems). PCR products amplified from genomic DNA containing geneΔ::KanMX were individually transformed into yeast cells. They were confirmed by PCR.
Patch mating assays were performed as described previously (Donze et al., 1999). Cells that were transformed by GBD or GBD-fused genes were grown on selective medium (YMD) lacking tryptophan at 30°C for 3 days. Single colonies were spread on YMD plates containing mating tester lawns JRY19a with an OD600 of 2 and leucine and adenine for diploid colonies. The plates were incubated at 30°C for 3 days, and were photographed daily.
Fub1p-3xFLAG purification was performed as described previously (Iida and Araki, 2004), with some modifications. The cells were grown in YPD medium to an OD600 of 2, washed once with PBS buffer and resuspended in 0.8 ml lysis buffer [50 mM HEPES-KOH (pH 7.5), 300 mM KCl, 0.05% Tween-20, 0.005% NP-40, 10% glycerol, 0.2 μg/ml pepstatin A, 0.2 μg/ml leupeptin, 1 mM PMSF, 2 mM β-glycerophosphate, 2 mM NaF, 0.4 mM Na3VO4, 0.5 mM Na-pyrophosphate]. Cells were disrupted by glass beads with vortex mixer. Cell lysates were cleared by centrifugation for 20 min at 4°C. Protein extracts were adsorbed onto 0.5 ml of sepharose 4B Fast Flow (GE Healthcare) for 1 hour at 4°C. Then, the beads were pelleted and the supernatant was recovered and mixed with 50 μl anti-Flag-agarose (M2) beads (Sigma) for 3 hours at 4°C. The beads were washed six times with 1 ml cold lysis buffer containing 0.1 mg/ml BSA, and then washed five times with 1 ml cold lysis buffer. The complex was eluted by incubating the beads with lysis buffer containing 100 μg/ml 3X FLAG peptide (Sigma).
Affinity-purified proteins were resolved by 15% SDS-PAGE. After silver staining with SilverQuestTM Silver Staining Kit (Invitrogen), the peptide bands were excised from the gel and subjected to in-gel reduction with 10 mM DTT, alkylation with 55 mM iodoacetamide, and digestion with 10 μg/ml modified trypsin (Promega) at 37°C for 16 hours. After in-gel digestion, the collected peptides were subjected to mass spectrometry analysis as described previously (Sadaie et al., 2008).
A total 55 genes related to silencing region control were isolated from approximately 6,000 genes in S. cerevisiae by genome-wide screening (Oki et al., 2004). We focused on YCR076C/FUB1 whose biological function was still unknown.
To investigate the function of Fub1p, we used a FLAG purification system to isolate Fub1p binding proteins. Because tag-fused proteins sometimes lose activity, we assessed the activity of a Fub1p-3xFLAG fusion protein by the same method used for genome-wide screening (Oki et al., 2004). Our results showed that Fub1p retained function even when fused to a 3xFLAG at the C terminus (data not shown). Therefore, we used this construct to generate the yeast strain, FUY95. The Fub1p-3xFLAG protein was purified using anti-FLAG antibodies and eluted using FLAG peptides. Following SDS-PAGE and silver staining, bands were excised for LC-MS/MS analysis to identify proteins bound to Fub1p. The results identified a number of 20S CP core-associated factors, including β7/Pre4p, α6/Pre5p, α4/Pre6p, β4/Pre7p, α2/Pre8p and α3/Pre9p (Fig. 1A), all of which belong to the α or β subunit of the 20S CP (Fig. 1B).
 View Details | Fig. 1 Purification of Fub1p-interacting proteins and identification of 20S CP components. (A) Silver-stained gels show purification of Fub1p-3xFLAG and an untagged control strain. The approximate position of the Fub1p complex components identified by mass spectrometry is indicated. All proteins belong to the 20S CP of the proteasome. The asterisk indicates an immunoglobulin. Line 1 is the untagged control (FUY31). Line 2 is Fub1p-3xFLAG (FUY95). (B) Molecular structure of the proteasome. The 26S proteasome is composed of more than 33 different subunits, containing one 20S CP structure and two 19S RPs. The shadow box shows active subunits isolated from Fub1p-3xFLAG. |
Because Fub1p was shown to bind to proteasome particles (Fig. 1A), a genetic approach was used to clarify the mechanism of interaction. To examine the genetic interaction, we utilized target gene overexpression and deletion analysis. Most proteasome components are essential genes, so we chose well-characterized temperature sensitive (ts) mutants. Some components of the 20S CP were isolated by LC-MS/MS, but these components were not proteolytically active. Among the 20S CP components, β1/Pre3p, β2/Pup1p and β5/Pre2p are known to possess proteolytic activities, thus, we chose the pre2-75 temperature sensitive mutant. On the other hand, the 19S RP was not isolated by our assay. Therefore, we tested the genetic interaction in rpt1 (cim5-1), rpt6 (cim3-1), rpt4-G106D and rpn1-821 temperature sensitive mutants. First, we observed the effects of overexpression of Fub1p in both wild-type (WT) and proteasome mutants. When Fub1p was overexpressed in WT, the cells exhibited obvious growth defects. Furthermore, Fub1p overexpression in pre2-75 (20S CP) mutant cells and rpn1-821, rpt1 (cim5-1) (19S RP) mutants resulted in much more serious growth defects than were seen in WT. However, there was no obvious effect of Fub1p overexpression in the rpt4-G106D, rpt6 (cim3-1) (19S RP) mutant cells (Fig. 2A).
 View Details | Fig. 2 Characterization of the genetic interactions between Fub1p and the proteasome complex components. (A) Effect of Fub1p overexpression on yeast growth. Control (p426GAL) and Fub1 overexpression plasmids (p426GAL-FUB1) were introduced in wild-type (WT, YPH499) and the proteasome temperature sensitive (ts) mutant strains, pre2-75 (des3), rpt4-G106D (des5), rpn1-821 (des16), rpt6 (cim3-1) (MHY4464) and rpt1 (cim5-1) (MHY4466). All plates were YM+galactose plates lacking uracil. Yeast were grown for 3 days at 30°C. (B) The fub1Δ rpt1 (cim5-1) double mutant showed a synthetic sick growth phenotype. WT (YPH499), proteasome ts mutants (rpt1 (cim5-1), pre2-75, rpt4-G106D, rpn1-821 and rpt6 (cim3-1), fub1Δ (FUY567), and a double mutant of fub1Δ and proteasome ts mutants were streaked on YPD plates and grown for 3 days at 30°C, 37°C and 38°C. |
Next, we analyzed the effects of fub1 deletion because no phenotype had been reported between fub1Δ and proteasome components. A double mutant of fub1Δ and rpt1 (cim5-1), a temperature sensitive mutant of the 19S RP, exhibited a synthetic sick growth phenotype. However, genetic interaction with Fub1p was not observed in pre2-75, rpt4-G106D, rpt6 (cim3-1) or rpn1-821 mutants (Fig. 2B). Based on these results, Fub1p interacts not only with the 20S CP but also with the 19S RP.
To determine whether Fub1p could maintain its boundary function in proteasome mutant strains, analysis was performed using the same method as the patch mating assay with genome-wide boundary screening (Oki et al., 2004). First, a plasmid (pFOM57), Gal4-binding domain (GBD) fused to full-length Fub1p, was introduced into wild-type (WT) and proteasome component deleted strains that had a Gal4-binding sequence adjacent to the silencing domain at HMR. If the Fub1p maintained boundary function, expression of the reporter gene a1 would be de-repressed and the cells would be unable to form diploid colonies on selective plates. Conversely, if cells did form diploid colonies on selective plates, this would indicate that the reporter a1 gene was repressed and that Fub1p had lost boundary function. Of the proteins isolated by LC-MS/MS, only Pre9p (a component of the 20S CP) was a non-essential gene (Hochstrasser, 1996), and Fub1p boundary function was lost in the pre9 mutant strain (Fig. 3). This result suggested that the 20S CP is important for Fub1p boundary function. Next, we tested whether the ump1 mutant strain would affect the 20S CP since Ump1p plays a role in assembly of the 20S CP (Bochtler et al., 1999; Collins et al., 2007; Maurizi, 1998; Ramos et al., 1998). As shown in Fig. 3, the ump1 mutant also lost boundary function of Fub1p. As a control, Sas2p, the histone acetyltransferase (HAT) catalytic subunit of the SAS-I complex, maintained its boundary function in all mutant strains as well as in the WT strain. These results suggested that Fub1p cooperated with Pre9p and Ump1p to regulate transcriptional silencing. Meanwhile, mutants of the 19S RP components rpn10 and rpn13, whose products recognize and bind to polyubiquitin chains (Husnjak et al., 2008), showed no change in the barrier activity of Fub1p. We also tested the hsm3 mutant, which is defective in base assembly, and observed the same result (Funakoshi et al., 2009; Roelofs et al., 2009; Saeki et al., 2009). Based on these results, we hypothesized that the 19S RP was unnecessary for the boundary function of Fub1p. We also observed loss of Blm10p because Blm10p is a multifunctional protein similar to mammalian PA200, which functions as an ATP-independent 20S CP activator associated with 20S CP assembly (Doherty et al., 2004; Fehlker et al., 2003; Schmidt et al., 2005; Stadtmueller and Hill, 2011). However, our results revealed no effect on the boundary function of Fub1p.
 View Details | Fig. 3 Boundary activity in mutant strains. Patch mating assay is shown with WT (ROY2042), pre9Δ (FUY563), ump1Δ (FUY312), rpn10Δ (FUY310), rpn13Δ (FUY311), hsm3Δ (FUY309) and blm10Δ (FUY375) strains to detect genes necessary for the boundary function of Fub1p or Sas2p. GBD is used as boundary formational negative control and GBD-Sas2p is used as positive control. |
To determine whether other boundary proteins isolated by genome-wide screening (Oki et al., 2004) were involved in boundary establishment by Fub1p, the boundary activities of some of these proteins were analyzed in a fub1-deleted strain. The proteins selected for analysis were related to the histone modification complex (Sas2p, Sas5p, Dot1p, Ada1p and Ada2p), TFIID complex (Taf47p), chromatin remodeling complex (Snf6p), mediator (Med2p), and cell cycle (Clb1p), as well as other factors such as Gic1p and Gds1p. The results showed that they did not appear to have any effect on boundary function of Fub1p (data not shown). Based on this finding, it appears that Fub1p functions independently, rather than cooperatively with the other boundary factors.
To analyze which domain of Fub1p is crucial for boundary function, a minimum domain analysis was performed (Fig. 4). The domain containing amino acids 165–229 of Fub1p is reported to be the PI31 proteasome regulator domain conserved in higher eukaryotes (Zaiss et al., 2002). In vitro, PSMF1, a human protein containing the PI31 proteasome regulator domain, behaves as a proteasome inhibitor (Mccutchen-Maloney et al., 2000; Zaiss et al., 1999, 2002). Patch mating assay results showed that the Fub1p domain containing amino acids 165–250 retained Fub1p boundary function (pFOM386, 165–250 aa). When the PI31 proteasome regulator domain was divided into two fragments (pFOM391, 196–250 aa; pFOM392, 165–195 aa), both maintained boundary function, but pFOM392 was not as strong as that of pFOM391. However, boundary function was lost when the C terminal domain (pFOM395, 196–237 aa), which has low homology to other species, was deleted.
 View Details | Fig. 4 Domain analysis of Fub1p boundary activity. Schematic presentation of Fub1p domain analyses that were tested with patch mating assay. (–) and (+) indicate constructs without and with the Gal4p binding sequence, respectively. The full length of Fub1p is shown as 1–250 aa and the black box (165–229 aa) indicates the PI31 proteasome regulator domain. |
The PI31 proteasome regulator domain in Fub1p where the boundary function is maintained is widely conserved from yeast to mammals (Fig. 5A, (Zaiss et al., 2002), http://www.ebi.ac.uk/interpro/IEntry?ac=IPR013886). Fub1p and PSMF1 are 19.6% homologous (similarity 32.2%) at the full length and 30.8% homologous (similarity 46.2%) in the PI31 proteasome regulator domain. Because there was high homology between Fub1p and PSMF1, we tested whether the boundary function of Fub1p was conserved in PSMF1. To test boundary function by patch mating assay, we cloned PSMF1 from a human cDNA library and constructed a GBD-PSMF1 fusion plasmid. To prevent false positive results due to the formation of heterodimers between GBD-PSMF1 and endogenous yeast Fub1p, GBD-fused plasmids (GBD, GBD-Fub1p and GBD-PSMF1) were introduced in both wild-type (WT) and fub1Δ mutant strains. As shown in Fig. 5B, PSMF1 maintained the boundary function at approximately the same level as Fub1p in both wild-type yeast and the fub1Δ mutant strains.
 View Details | Fig. 5 Boundary activity of the human PSMF1. (A) Sequence alignment of S. cerevisiae Fub1p and Homo sapiens PSMF1/PI31. Stretcher (Version 6.3.1; http://www.ebi.ac.uk/Tools/psa/emboss_stretcher/) was used for alignments. A line (|) at the bottom of the column of aligned amino acids indicates 100% identity, two dots (:) indicate similar amino acids. (B) H. sapiens PSMF1 also had boundary function. Wild-type (ROY2041 and ROY2042) and fub1Δ mutant strains (FUY165 and FUY166) were transformed with GBD, GBD-Fub1p or GBD-PSMF1 plasmids, and patch mating assays were performed as described previously. (–) and (+) indicate constructs without and with Gal4p-binding sites, respectively. |
To investigate the in vivo role of Fub1p in the control of transcriptional silencing, we performed LC-MS/MS. Although Fub1p was previously reported to interact with the proteasome composition factors Pre2p and Pup3p (Collins et al., 2007; Krogan et al., 2006), this association had yet to be analyzed in depth. To determine whether Fub1p interacted with the 20S CP, the 26S proteasome consisting of both 19S RP and 20S CP, or just Pre2p and Pup3p, we purified the interacting proteins using an Fub1p-3xFLAG strain. The results demonstrated that Fub1p bound to subunits of the 20S CP; however, subunits of the 19S RP were not detected. Based on in vitro analysis with the PI31 proteasome regulator domain of PSMF1, which is highly homologous to the Fub1p C terminal region, the PI31 proteasome regulator domain appeared to bind to the α ring of the 20S CP and inhibit its activity (Chu-Ping et al., 1992; Mccutchen-Maloney et al., 2000). To analyze the genetic interaction of Fub1p with the proteasome, growth defects were assessed following overexpression of Fub1p. These defects are thought to be caused by inhibition of the enzymatic activity of the 20S CP or by deregulation of the ubiquitin-proteasome system. The slow growth of the pre2-75 mutant strain was likely due to reduced enzymatic activity of the 20S CP, while the more severe growth defects seen in the rpn1-821 and rpt1 (cim5-1) mutant strains were probably caused by a decreased unfolding ability of the subunits of the 19S RP, which reduced the efficiency of translocation into the central channel of the 20S CP. On the other hand, the 19S RP mutant strains rpt4-G106D and rpt6 (cim3-1) did not exhibit an obvious growth defect phenotype. We believe the reason for this is that another 20S CP activator, such as Blm10p/PA200 (Yeast/Human) or PA26/PA28 (Yeast/Human), was able to compensate for 19S RP activity in these strains (Stadtmueller and Hill, 2011). We isolated the 20S CP factor that interacted with the Fub1p by affinity purification. Furthermore, in genetic interaction, both the 20S CP factor and the 19S RP factor were detected. Although this appears to contradict our earlier results, it is possible that the 19S was not detectable under the conditions of the affinity purification experiment. Direct interaction between the 19S RP and Fub1p might be observed by changing the experimental conditions.
When the boundary function in our focused silencing domain was examined, Fub1p lost boundary function in cells in which pre9 was deleted, the only single factor that can be disrupted among all the subunits of the 20S CP. Fub1p boundary function was also lost in cells lacking ump1, a chaperone of the 20S CP. These results suggested that the entire 20S CP plays an important role in boundary formation for controlling the spread of transcriptional silencing. On the other hand, when all the factors of the 20S CP α ring were fused with the GBD, no boundary function was observed using the same methods as in Fig. 3 and Fig. 4 (data not shown). Since the GBD was fused to the amino terminus of the protein in our system, it is possible that the GBD-20S CP did not maintain normal function. Conversely, there was no loss of Fub1p boundary function in the 19S RP mutant strains, indicating that the 19S RP might not be necessary for direct boundary function, or that 19S RP function could be compensated for by other 20S CP activators, such as Blm10p/PA200 and PA26/PA28 (Stadtmueller and Hill, 2011).
It was very important to clarify the native barrier function of Fub1p. Therefore, we generated a Fub1p-3xFLAG strain and investigated whether Fub1p was present at the native boundary region using ChIP assay with anti-FLAG antibody. In budding yeast, a native barrier is located in the telomeric region of each chromosome, HML, HMR and an rDNA coding region, and we examined the left side boundary region at HMR. However, no Fub1p was observed in this region (data not shown). To determine whether the silencing region would extend to the left side in a fub1Δ mutant strain considering the loss of the left side boundary at HMR, we used ChIP assay with anti-Sir3p antibody, and no extension was found (data not shown). It was also possible that Fub1p participated in the barrier function at other silencing regions, such as the telomeres, HML, and the rDNA coding region. As other possibility, no obvious change was observed in the boundary function or extension of the silencing domain by removing either of the two elements, the cis-element and HAT (a histone acetylation enzyme complex), that participated in the function of the HMR right side boundary (Oki and Kamakaka, 2005).
Many reports examine the relationship between chromatin and the proteasome. For example, it is reported that the proteasome interacts with chromatin directly (Sikder et al., 2006), the composition factor of the proteasome is required for extension in RNA transcription (Bhaumik and Malik, 2008; Ferdous et al., 2001), proteasome works on the chromatin (Szutorisz et al., 2006). In addition, ts mutations in both the 19S RP and 20S CP proteasome resulted in the loss of silencing activity (Ezhkova and Tansey, 2004). Furthermore, double mutation of sem1 and ubp6, a composition factor of the 19S RP and deubiquitinating protein, respectively, weakened the effect of Sir proteins at the telomere silencing domain (Qin et al., 2009). Similarly, a similar result was observed at the telomere domain following loss of rpn10 (Hang and Smith, 2011). These data also supported that the proteasome might affect the boundary function. From the above findings, we hypothesize that Fub1p interacts with the proteasome on chromatin and forms a boundary either by making use of the activity of the 20S CP or by weakening silencing through inhibiting proteasome function.
The PI31 proteasome regulator domain lies in the C terminus of Fub1p. Fub1p possesses two boundary formation domains, as shown by the results of the minimum domain analysis (Fig. 4). Both of these domains retained boundary function, even in the absence of the full PI31 proteasome regulator domain, although pFOM392 was not as strong as either pFOM391 or the full Fub1 domain. Based on these findings, it is plausible that both or one of these domains binds the 20S CP. In fact, interaction between the PI31 proteasome regulator domain of PSMF1 and the 20S CP has already been reported (Mccutchen-Maloney et al., 2000; Zaiss et al., 1999, 2002). Therefore, it will be interesting to ascertain whether the other minimum domain can also bind the 20S CP or other factors, and to identify any additional factors that interact with the N terminus of Fub1p.
It is thought that human PSMF1 may be an ortholog of Fub1p based on conservation of the PI31 proteasome regulator domain at the C terminus of Fub1p. In addition, as the proteasome is present in species from bacteria to the eukaryotes, it is tempting to speculate that Fub1p may have a similar boundary function in other species. However, because the mechanism by which PSMF1 might participate in the process boundary formation to restrict the spread of silencing in human cells is still unknown, future studies are needed.
We thank Drs. Kamakaka and Tsukiyama for providing yeast strains and plasmids. This work was supported by the JST PRESTO program and by a grant from the Research and Education Program for Life Science.
|
Index