Transcriptional profiling of ubp10 null mutant reveals altered subtelomeric gene expression and insurgence of oxidative stress response.

UBP10 codes for a deubiquitinating enzyme of Saccharomyces cerevisiae whose loss of function determines slow growth rate and partial impairment of silencing at telomeres and HM loci. A genome-wide analysis performed on a ubp10 disruptant revealed alterations in expression of subtelomeric genes together with a broad change in the whole transcriptional profile, closely parallel to that induced by oxidative stress. This response was accompanied by intracellular accumulation of reactive oxygen species as well as by DNA fragmentation and phosphatidylserine externalization, two markers of apoptosis. SIR4 inactivation mitigated the wide transcriptome remodeling of the ubp10 null mutant affecting particularly the stress transcriptional profile. Moreover, the ubp10sir4 disruptant did not display apoptotic markers. These results argue in favor of an involvement of deubiquitination in transcriptional control and suggest a linkage between oxidative stress and apoptotic pathway in budding yeast.

UBP10 codes for a deubiquitinating enzyme of Saccharomyces cerevisiae whose loss of function determines slow growth rate and partial impairment of silencing at telomeres and HM loci. A genome-wide analysis performed on a ubp10 disruptant revealed alterations in expression of subtelomeric genes together with a broad change in the whole transcriptional profile, closely parallel to that induced by oxidative stress. This response was accompanied by intracellular accumulation of reactive oxygen species as well as by DNA fragmentation and phosphatidylserine externalization, two markers of apoptosis. SIR4 inactivation mitigated the wide transcriptome remodeling of the ubp10 null mutant affecting particularly the stress transcriptional profile. Moreover, the ubp10sir4 disruptant did not display apoptotic markers. These results argue in favor of an involvement of deubiquitination in transcriptional control and suggest a linkage between oxidative stress and apoptotic pathway in budding yeast.
Covalent attachment of ubiquitin (Ub) 1 to specific proteins represents a reversible post-translational modification associated not only with the removal of damaged or misfolded proteins but also involved in the regulation of different important cellular processes from signal transduction and cell cycle progression to endocytosis and apoptosis (1)(2)(3)(4)(5). The complexity of the Ub-dependent system reflects the wide range of its targets including cell cycle regulatory proteins like cyclins, oncogenes such as c-mos, c-jun, and c-fos, and tumor suppressor genes products such as p53. It is not surprising that perturbations in ubiquitination are implicated in the pathogenesis of several neoplastic and neurodegenerative diseases (6 -8).
Important for the efficiency of the Ub-dependent pathway is the activity of a family of deubiquitinating enzymes (Dubs) acting as specific thiol proteases. These enzymes hydrolyze the linkage between the C-terminal Gly-76 of Ub and a Lys residue of a given substrate or of another Ub molecule. Dubs represent the largest known group of enzymes in the Ub system, and they fall into two distinct types: Ub C-terminal hydrolases and Ubspecific processing proteases (Ubps). The former includes low molecular weight enzymes releasing Ub from small size substrates such as peptides or glutathione; the latter includes higher molecular weight enzymes removing Ub from different substrates, at least in vitro. Ubps are highly divergent, but they contain several short consensus sequences (Cys and His boxes) essential for their enzymatic activity (1). Despite the identification of a great number of Dubs, their physiological roles remain quite obscure. They are required at several levels along the Ub pathway playing different functions. Ub is always synthesized as a precursor requiring removal of C-terminal peptides or amino acids; so Dubs are needed to generate Ub monomers. Furthermore, activated Ub can form stable complexes with abundant intracellular nucleophiles, such as amines and glutathione. This reaction, if not balanced, leads to a depletion of the cellular Ub pool. Moreover, after proteasomedependent proteolysis of the substrate, released poly-Ub chains have to be rapidly disassembled. Finally, besides being involved in Ub recycling, Dubs play a regulatory function controlling ubiquitination level and activity of ubiquitinated proteins (9 -11).
After the complete sequencing of the Saccharomyces cerevisiae genome, 17 genes encoding Dubs have been identified, among which are 16 Ubps. Few Ubps have a known physiological function (12). For example, Ubp14 disassembles unanchored polyubiquitin chains into monomeric Ub (13), and Ubp4/ Doa4, similar to human oncogene tre-2 (14), plays an important role in Ub recycling from both proteasome and vacuole (15)(16)(17). In yeast, the genetic redundancy of Ubps suggests a high degree of substrate specificity and thus a direct involvement in the activity of specific proteins, whereas the fact that none of the different Ubps is encoded by an essential gene points to some overlapping functions. We focused our studies on the UBP10/DOT4 gene because of the intriguing ubp10 phenotype. First, cells lacking Ubp10p show a partial loss of silencing at telomeres and at the mating type (HM) loci (18), a reduction in Sir4p level in agreement with a Ubp10p nuclear localization and with its ability to interact with Sir4p in a two-hybrid assay (19). Second, ubp10 mutants display growth defects at different temperatures exacerbated when UBP10 is inactivated in strains with several auxotrophic markers (12). The slow growth rate can be partially complemented by mutations in a subunit of the 26 S proteasome (doa3-1) or by deletions of SIR2, SIR3, or SIR4 suggesting that Ubp10p activity is important for restricting silent information regulator (Sir) proteins to the normal silent loci by the regulation of Sir4p (19). In S. cerevisiae the products of the four SIR genes are important factors for the establishment and maintenance of a specialized chromatin structure, called silent chromatin, analogous to heterochromatin of higher eukaryotes. Silent chromatin is responsible for silencing at the HM loci, telomeres, and rDNA repeats. All the Sir proteins contribute to silencing the HM loci, whereas silencing at telomeres requires Sir2p, Sir3p, and Sir4p (reviewed in Refs. 20 and 21). In addition, Sir2p, an NAD ϩ -dependent deacetylase, is involved in rDNA silencing (reviewed in Ref. 22). Sir proteins are recruited to DNA through a network of protein-protein interactions involving Sir proteins themselves and other factors including Rap1, Abf1, the origin recognition complex, and histones H3 and H4. The silencing efficiency is linked to the relative concentration of Sir proteins due to competition among the different silent loci for limited Sir proteins. In this regard, deletion of SIR4 severely reduces telomeric and HM silencing, preventing Sir3p recruitment to the telomeres (23), and enhances rDNA silencing indicating that telomeres and rDNA repeats compete for a limited amount of Sir2p (24). Conversely, SIR4 overexpression disrupts silencing suggesting that Sir4p overexpression may titrate Sir3p away from chromatin and Sir2p from the rDNA (25). In fact, Sir4p is a large multidomain protein that is able to interact with many other proteins such as Sir2p, Sir3p, Sir4p, Rap1p, and yKu70p (20,26). In particular Sir4p binding to Rap1p initiates the sequential assembly of the Sir complex at telomeric ends allowing the subsequent spreading along the chromosome (27). Moreover, it has also been reported that the C-terminal domain of Sir4p can associate with Ubp3p (28) and Ubp10p (19), suggesting that ubiquitination could target other proteins of the silencing machinery in addition to histones.
However, the failure in detecting ubiquitinated Sir4p (19) together with the partial complementation of the ubp10 growth defect observed by inhibition of endocytosis (29) suggest that precise functions and additional targets of Ubp10p remain to be discovered.
In this report we integrated a genome-wide approach with phenotypic analyses to dissect mechanisms underlying Ubp10p function. Because Ub is involved in many physiological processes and the ubp10 phenotype indicates perturbations in different cellular functions, we thought that a global transcriptional profiling could be useful for extracting functional information to be verified by in vivo assays. Our analysis revealed that deletion of UBP10 induced a huge transcriptome remodeling characterized by a reduced subtelomeric repression and up-regulation of stress-responsive genes. This was accompanied by accumulation of reactive oxygen species (ROS) and by appearance of apoptosis-like phenotypical markers. Following SIR4 inactivation, these features were completely abrogated except for transcriptional derepression of silencing that partly reflects SIR4 loss of function. These results further support that the Sir-mediated activity is influenced by (de)ubiquitindependent signaling mechanisms, and it is the first indication that in yeast a deubiquitinating enzyme may be involved in programmed cell death.
Recombinant DNA Procedures and Plasmids-Standard protocols were used for recombinant DNA manipulation and yeast transformation (32,33). Full-length UBP10 and SIR4 genes were amplified from yeast chromosomal DNA with the following primers: UBP10-up (5Ј-CAGAGTCAGAGTGGCGCACTA-3Ј) and UBP10-down (5Ј-CGCATT-GGGCTCCAAGGGTAT-3Ј), SIR4-up (5Ј-ACCGGTATCTGGGCTGGT-GTTGA-3Ј) and SIR4-down (5Ј-CATCCCTGCAGCTGTCCGAACAA-3Ј). The two PCR products, containing the UBP10 coding sequence flanked by 561 bases upstream and 718 bases downstream and SIR4, spanning between Ϫ523 and 5071 nucleotides from the ATG, were subcloned into the pCR®-Blunt vector (Invitrogen) to produce pUBP10 and pSIR4 plasmids, respectively. PCR products were routinely checked by sequence analysis. To generate ubp10::HIS3 cassette, a BamHI/XhoI fragment containing the HIS3 gene was inserted by blunt-end ligation into pUBP10 cut with ClaI and HpaI. Blunt-end cloning of a 2.2-kb SalI/ XhoI LEU2 fragment into ScaI/ClaI-digested pSIR4 replaces the SIR4 ORF. Gene disruptions were confirmed by diagnostic PCR and Southern analyses. Gene Chip® Analysis-Cells for RNA isolation were rapidly collected by filtration, and filters were frozen at Ϫ80°C. Total RNA was extracted from frozen cells according to the method of selective precipitation with LiCl (34). Fragmented antisense cRNA was prepared following Affymetrix (Santa Clara, CA) recommendations. Briefly, doublestranded cDNA was retro-transcribed from 30 g of total RNA using a modified oligo(dT) primer with a 5Ј T7 RNA polymerase promoter sequence and the Superscript Choice System for cDNA synthesis (Invitrogen). cRNA, obtained using an ENZO kit (Affymetrix), was purified on an affinity column (RNeasy, Qiagen). Analysis was done by hybridizing the fragmented cRNAs to the Affymetrix GeneChip® Yeast Genome S98 array, which permits the monitoring of the mRNA abundance from 6400 ORFs. Probe array hybridizations were carried out under rotation at 45°C for 16 h as described (35). The arrays, stained by incubation with 2 g/ml phycoerythrin-streptavidin (Molecular Probes, Eugene, OR) and 1 mg/ml acetylated bovine serum albumin (Sigma), were read at a resolution of 7.5 m by a confocal scanner and analyzed with the MicroArray Suite 4.0 Gene Expression analysis program (both from Affymetrix). The whole set of data is available at www.ncbi.nlm. nih.gov/geo/(GEO series accession number GSE804).
Reverse Transcription (RT)-PCR-Total RNA, prepared as described above for Gene Chip analysis, was purified using the RNeasy RNA purification kit and then treated with DNase I, RNase-free (Roche Applied Science), for 1 h at 37°C followed by phenol extraction and ethanol precipitation. RT-PCR was carried out to amplify the ARO9, AUT4, CTT1, GLK1, GPD1, HSP26, HXK1, and ACT1 mRNAs using the Access RT-PCR System (Promega) following the manufacturer's instructions. The number of cycles was lowered to 15/20 so that amplification was in the exponential range. Experiments were repeated at least twice with different RNA preparations. Primers sequences used for PCRs are available upon request.
Northern Analysis-Northern analysis was performed as reported previously, using 32 P-radiolabeled RNA probes generated by in vitro transcription (36). ACT1 mRNA was used as an internal standard.
Test for Apoptotic Markers-Free intracellular radicals were detected with dihydrorhodamine 123 or dichlorodihydrofluorescein diacetate (dichlorofluorescein diacetate; Sigma) as described previously (37). For flow cytometric analysis, cells were incubated with dichlorofluorescein diacetate for 2 h and analyzed using a FACS® Star (BD Biosciences) with excitation and emission settings of 488 and 525-550 nm (filter FL1), respectively. TdT-mediated dUTP nick end labeling (TUNEL) test was performed according to Ref. 37, except for spheroplast preparations carried out with Zymolase 100T (ICN Biomedicals). DNA ends were labeled using the In Situ Cell Death Detection Kit, POD (Roche Applied Science). Cells were examined under light microscope and through a fluorescein optical filter. Externalization of phosphatidylserine was detected essentially as described (38). Spheroplasts were examined under fluorescence microscope after 20 min of incubation at room temperature with FITC-labeled annexin V (ApoAlert Annexin V Apoptosis Kit; Clontech, Palo Alto, CA) and propidium iodide (50 g/ml).

RESULTS AND DISCUSSION
Transcriptome Profiling of ubp10 Null Mutant Reveals a Cellular Stress Response-ubp10 mutants have a detectable phenotype characterized by an impairment of silencing at telomeres and HM loci and a reduced growth partially complemented by SIR4 inactivation (12, 18, 19). We generated a set of null mutants carrying single or double disruptions (ubp10, sir4, ubp10 sir4) in the wild type backgrounds W303-1A (mating  TABLE I Group of 82 stress-related genes that display differential (increased) expression between ubp10⌬ and W303-1A strains Genes are divided into 11 functional sub-groups St, genes containing in the promoter region one or more stress-responsive elements (STRE); Et, genes whose expression usually increased after a treatment with high ethanol concentrations; H 2 O 2 , genes whose expression usually increased upon hydrogen peroxide stimulus; Os, genes whose expression usually increased in hyperosmotic stress conditions. type a) and -1B (mating type ␣) whose phenotypes were similar to published ones (12,19,39).
To get insight into the cellular functions affected by UBP10 inactivation, we first carried out a microarray analysis of the differential genome-wide expression profile of a ubp10 mutant versus its isogenic wild type strain (W303-1A). Data from two independent pairs of experiments were analyzed. Cultures were harvested at a cellular density of 5 ϫ 10 6 /ml in exponential growth phase on 2% glucose/YNB medium, a growth condition in which the ubp10 mutant showed a more severe phenotype. Transcript levels obtained from the two different strains were compared using the Affymetrix method of scaling. Reliability of the gene chip analysis was also evaluated by comparing transcript levels of three reference genes, ACT1, PDA1, and HHO1, commonly used as standards (40).
Comparison between ubp10 null mutant and wild type strain revealed changes in transcript levels of a large fraction of the genome. Such remodeling of the transcriptome reflects perturbations of different cellular functions, and it can correlate with some phenotypic traits of ubp10 cells, in particular with its involvement in telomeric silencing and with the requirement of Sir4p in maintaining proper chromatin function. Given the amount of information generated, only some relevant aspects will be presented in this report; the full data set is available at www.ncbi.nlm.nih.gov/geo/. Initially, genes with a known cellular role were clustered according to their physiological path-way. Both induced and repressed genes belonged to very heterogeneous functional categories; however, we found some classes to be more represented than others. The down-regulated group included genes coding for the following: (i) various transporters, such as tyrosine, glutamate, isoleucine permease, ammonium permease, biotin, and peptides carriers; (ii) ribosomal proteins and translational factors; and (iii) enzymes involved in the biosynthesis of arginine, lysine, tryptophan, and nucleotides. The down-regulation of some of these mRNAs could indicate an impairment in protein synthesis and nutrient transport related to the ubp10 disruptant growth defect that is particularly evident when the deletion is made in strains with auxotrophic markers (especially his3, leu2, lys2, and trp1) (19). No change was observed in GAP1 transcript in agreement with Northern analysis (29).
Among the genes whose expression was increased, a conspicuous group encoded the following: (i) enzymes involved in carbohydrate transport, metabolism, and energy generation; (ii) heat shock proteins; and (iii) transcription factors, such as Hap2, Hap4, Rox1, Yap4. Because many of these have been found previously to be down-or up-regulated in specific physiological conditions, we analyzed our results not only according to metabolic pathways but also on the basis of published data (41)(42)(43)(44). This further clustering revealed that many of the genes induced in the ubp10 null mutant had been proposed previously to offer cell protection in response to stress, i.e.

TABLE II
Group of 24 subtelomeric genes, which display differential expression between upb10⌬ and W303-1A strains hyperosmotic or oxidative stress, heat shock, ethanol exposure, and starvation. These genes are listed in Table I depending on functional categories. In addition, we found a subset of 25 genes that were characterized by the presence of stress-responsive consensus sequences in their promoter regions. Transcription driven by stress-responsive elements through Msn2 and Msn4 factors has been demonstrated to be induced by a variety of stress conditions activating the general stress response (44). Among the 25 genes there was some coding for enzymes that synthesize trehalose (TPS1, TPS2, and TSL1), glycogen (GSY1, GSY2), and their precursors, as well as others encoding enzymes involved in the degradation of carbohydrates (ATH1,  NTH1, NTH2, and GPH1). The simultaneous induction of both synthetic and catabolic enzymes allows cells to rapidly buffer and manage osmotic instability and energy reserves. In particular, trehalose protects cellular components from detrimental effects of stress by providing the energy required for renaturation of cellular structures and protecting cells and membranes from denaturation (42,43,45). Other genes involved in carbohydrate metabolism were GPD1 and GPP2 encoding glycerol-3-phosphate dehydrogenase and glycerol-1-phosphatase, respectively, two glycerol-producing enzymes that play a critical role in response to the hyperosmotic shock allowing the accumulation of glycerol as compatible osmolite (46).
Nine genes (HSP26, HSP42, HSP78, HSP104, HSP12, YRO1, YRO2, SSE2, and SSA4) encoding heat shock proteins were strongly up-regulated (Table I). Hsp104 is a member of the highly conserved Hsp100/Clp family and acts as a chaperone to disaggregate damaged proteins. SSA4 encodes the cytosolic Hsp70 that prevents protein aggregation directly, and it is required for protein refolding together with Hsp40 and Hsp104. In addition, among the functions related to the metabolism of damaged proteins, in ubp10 null mutant cells we found an induction of the polyubiquitin gene UBI4 whose expression is normally enhanced under conditions of heat shock, starvation, respiratory growth or other conditions where damaged or partially denatured proteins need to be degraded to prevent their accumulation as aggregates (47). A vacuole-mediated process FIG. 1. Validation of microarray data. A, Northern analysis was performed on total RNA isolated from the ubp10⌬ mutant and its isogenic wild type. About the same amount of RNA was loaded on each lane (20 g) and hybridized with probes specific for the indicated genes. ACT1 mRNA is shown as the loading control for each hybridization. B, the expression of ARO9, AUT4, CTT1, GLK1, GPD1, HSP26, and HXK1 in the wild type (W303-1A) and ubp10⌬ strains was analyzed by semiquantitative RT-PCR as described under "Experimental Procedures." ACT1 was used as a control. Similar data were obtained for RNA independent samples.

FIG. 2. SIR4 disruption suppresses the induction of the stress-related genes in the ubp10⌬ mutant.
A, fold change values plot referred to a group of 156 stress-related genes, which display differential expression between ubp10⌬ versus W303-1A (E), ubp10⌬sir 4 ⌬ versus W303-1A (OE), and ubp10⌬sir4⌬ versus ubp10⌬ (Ⅺ). Values referred to ARO10, YJL144W, YRO1, and YRO2 genes were omitted for a graphical reason. B, fold change values plot referred to a subgroup of 69 stress-related genes, which display differential expression between ubp10⌬ versus W303-1A (E) and ubp10⌬sir4⌬ versus ubp10⌬ (q). Values referred to YRO1, YRO2, and YJL144W genes were omitted for the same reason as in A. Only genes with a fold change Ն͉2͉ are shown.

TABLE III
Group of 54 subtelomeric genes, which display differential expression between ubp10⌬sir4⌬ and W303-1A strains called autophagy also contributes in the removing of damaged cellular components (48); UBP10 loss of function induced the expression of genes involved in this process (APG7, AUT4, and AUT7) ( Table I).
Changes in the expression of components of the Ub system have been shown to take place in response to oxidative stress (49,50); a notable feature of the ubp10 mutant transcriptome was the induction of genes whose products act as scavengers in the enzymatic defense against oxidizing agents that potentially damage proteins and nucleic acids (51,52). We detected GRX1, encoding the cytoplasmic glutaredoxin; GPX1, glutathione peroxidase; SOD2, mitochondrial Mn-superoxide dismutase; CCP1, mitochondrial cytochrome c peroxidase, CTT1, cytoplasmic catalase T; and GTT1, glutathione transferase. In particular, the increased expression of SOD2, CCP1, and CTT1 is a typical phenomenon observed during respiratory metabolism to counteract ROS production (52-54).
To confirm the reliability of the microarray data, a subset of selected genes, belonging to different functional groups that exhibited small and large transcriptional changes, was further characterized by Northern analysis (Fig. 1A) and semiquantitative RT-PCR (Fig. 1B). As illustrated in Fig. 1, the results obtained with the two different approaches revealed the same pattern of induction for the nine genes analyzed as that shown by microarray analysis (Table I).
In conclusion, the lack of Ubp10p activity results in the activation of a defined program of gene expression that displays all the characteristics of an adaptive response induced by oxidative stress. In ubp10 mutants as in oxidatively stressed cells, there is an increased need for protection against a damage in cellular homeostasis that takes place without external perturbation but only upon the absence of a deubiquitinating activity.
SIR4 Inactivation Suppresses the Transcriptional Stress Response of the ubp10 Mutant-As a further refinement of our  Reduced Silencing and Oxidative Stress in ubp10 Mutant analysis, we examined the genome-wide transcriptional consequences of deleting SIR4 in a ubp10 background to see if ubp10⌬sir4⌬ transcriptional profiles had detectable features that correlated phenotypic and biochemical data (i.e. the partial complementation of the growth defect).
Evaluation of the ubp10 sir4 transcriptional profile revealed that SIR4 inactivation mitigated the wide transcriptome remodeling of the ubp10 mutant (see www.ncbi.nlm.nih.gov/ geo/). In order to delineate which set of genes was particularly affected in the absence of Sir4p, we initially focused on stress-responsive genes (41)(42)(43). Fig. 2A shows the fold change values referred to the stress-related genes that were up-and downregulated in ubp10⌬ and ubp10⌬sir4⌬. A clear reduction in the transcriptional response was evident indicating a partial suppression of the stress response induced by loss of UBP10 alone. No significant change in the transcript level of the same genes was detected in the sir4 null mutant (data not shown). Fig. 2A also reports the fold change values of the stress-responsive genes obtained from the comparison between ubp10sir4 mutant versus the ubp10 one. It is remarkable that the majority of these genes falls in the category of down-regulated genes. Thus, the partial complementation of the ubp10 phenotype following SIR4 inactivation seems to reflect a reversion of the transcriptional profile detected in the ubp10 cells. In particular, focusing on the fold change values of a subset of 69 stress-related genes that were all present in the comparisons between ubp10/wild type and ubp10sir4/ubp10, the resulting plot had a symmetrical arrangement with two specular profiles (Fig. 2B). A similar plot was obtained for the fold changes values referred to some of the ubp10 down-regulated genes encoding various transporters, ribosomal proteins, and enzymes for the biosynthesis of amino acids (data not shown), suggesting a recovery of the ubp10sir4 phenotype strictly linked to a transcriptome remodeling.
Ubp10p has been proposed previously to play a role in Sir4p regulation, and the ubp10 phenotypic alterations have been suggested to be a consequence of a promiscuous binding of Rap1p-Sir complexes to inappropriate loci leading to a decrease in silencing and transcription of some Rap1p-regulated genes (i.e. ribosomal genes). Upon SIR4 inactivation, the incorrect recruitment of the silencing apparatus and the deleterious gene repression should be prevented (19). Our data are in agreement with this model, supporting an involvement of Ubp10p in transcriptional regulation. In exponentially growing cells, Rap1p targets 294 loci and participates in the activation of 37% of all RNA polymerase II initiation events (55). This fact together with the Rap1p specificity for binding to intergenic sequences potentially acting as promoters (55) could account for the transcription profiling changes detected in the ubp10 disruptant.
Reduced Silencing Displayed by ubp10 Cells Correlates with an Altered Subtelomeric Gene Expression-Genome-wide technology is a powerful tool for analyzing extensively the transcription of ORFs mapping in subtelomeric chromosomal regions; it allowed us to study the silencing phenomenon without using reporter systems. Referring to the analyses on the ubp10 single mutant, transcription profiling revealed a predominant effect of reduction in the silencing phenomenon. In detail, we found 24 subtelomeric genes subjected to altered transcription, of which 18 resulted in an increase by a fold change ranging from 2.1 to 10.9, whereas the fold change values of the 6 genes whose expression was reduced were included between Ϫ3.5 and Ϫ2.3 (Table II). In the double mutant ubp10sir4, we detected 54 ORFs whose transcriptional levels changed significantly; 50 appeared more expressed by a fold change ranging from 2.1 to 48.3 and 4 decreased, displaying a fold change between Ϫ5.8 and Ϫ2.5 (Table III). Rap1p and Sir proteins are required for the silencing of genes located predominantly within 5-10 kb of the ends of the telomeres (56, 57) (Fig. 3A). When we examined how the fraction of derepressed subtelomeric genes in the ubp10 disruptant varied their distance from the chromosome ends, we found UBP10 deletion up-regulated genes prevalently within 5-10 kb as for the sir4 mutant, although the global effect of derepression was less pronounced (Fig. 3A). Interestingly, in the double disruptant the percentage of up-regulated subtelomeric genes was similar to sir4 mutant one, but an increase in transcriptional derepression within 15-20 kb of telomeres was observed (Fig. 3A) affecting a subtelomeric region where other factors (i.e. the histone methyltransferase Set1) play a distinct role from the Sir proteins to maintain silencing (58). Thus, it appears likely that the effects of UBP10 deletion on subtelomeric silencing are not exclusively mediated by Rap1-Sir complexes.
In addition, Rap-Sir complexes are required for silencing at the HM loci. The MAT locus codes for three gene regulatory proteins that together with a larger group of proteins encoded elsewhere in the genome are responsible for the three distinct types of yeast cells (a, ␣, and a/␣) (59,60). Consistent with its effect on artificially inserted reporters (19), we found that deletion of UBP10 in a W303-1A background (MATa) induced a slight derepression of both HMRA1 and HMLALPHA2 genes (Fig. 3B), encoding a1 and ␣2 corepressors, respectively. In the double disruptant, a transcriptional response similar to the one obtained for sir4⌬ one was present; HMRA1, HMRA2, as well as HMLALPHA2 were actively expressed (Fig. 3B). Consequently, we observed the related repression of the entire set of a-specific genes (i.e. MFA1, MFA2, STE2, STE6, BAR1, and AGA2) and of some haploid-specific genes, such as FAR1, STE4 and STE5. These cells, as expected, were ␣-factor-resistant (data not shown).
In summary, upon evaluating the global telomeric genes expression in the different strains, the double mutant displayed a more evident derepression effect than ubp10, comparable (not identical) to sir4 single mutant. These results seem to exclude a synergic interaction between UBP10 and SIR4 inactivations, but point to an involvement in the same pathway, where Sir4p plays a main role or, alternatively, to parallel pathways with similar targets. ubp10 Cells Contain ROS in the Absence of External Oxidative Stress-Because the transcriptional stress response was a feature, emerging from the genome-wide analysis of ubp10, unrelated to phenotypic traits previously observed, we assayed the sensitivity of ubp10 cells to different stress conditions (high ethanol, dithiothreitol, UV, high osmolarity, and H 2 O 2 ). A plating assay revealed that mutant cells were more resistant than the wild type to increasing concentrations up to a sub-lethal dose of H 2 O 2 , suggesting an endogenous tolerance (data not shown). In particular, H 2 O 2 tolerance depends on catalase activity, whose gene was up-regulated in the ubp10 mutant (Table I). Such an endogenous oxidative condition occurs in the cells in the absence of any external oxygen radical sources when ROS, side products of the respiratory metabolism, are no longer detoxified (52,54). Mitochondrial accumulation of strongly oxidizing molecules is detectable using in vivo specific staining reactions depending on the oxidation of fluorochrome precursors. Both ubp10 and wild type cells in late exponential phase of growth on glucose were incubated with dichlorodihydrofluorescein diacetate for 2 h and then examined under a fluorescence microscope; the mutant population exhibited fluorescent dichlorofluorescein (about 12%), whereas the corresponding wild type cells did not show any signal (Fig. 4A). Viability staining with FUN-1 and Calcofluor White M2R revealed that the percentage of fluorescent dichlorofluorescein cells was not due to nonviable cells, and it corresponded to old mother cells (data not shown). The strain background W303 has been reported to harbor the rad5-535 allele, a weak allele of RAD5, which can have additional phenotypic effects when combined with other mutations (61). To rule out the possibility of synergist effects between this allele and ubp10 null mutation, we inactivated UBP10 in another strain, BY4741. The disruptants displayed growth defects similar to ubp10⌬ mutants in the W303 background (data not shown). Moreover, flow cytometric analysis of rhodamine 123 distributions (Fig. 4B) showed the presence of cells (about 17%) with high fluorescence in the ubp10 null mutant (BY4741 background) grown on glucose indicating that ROS accumulation was not strictly correlated to the rad5 mutation. Thus, the loss of Ubp10 activity triggers ROS accumulation in growth conditions that normally induce no increase of radical production. Finally, ROS presence was tested in cells growing on a non-fermentable carbon source, which imposes respiratory metabolism and consequently forces the mitochondrial activity. Dichlorodihydrofluorescein diacetate was added to ubp10, ubp10sir4, and wild type cells in late exponential phase of growth on glycerol; after 2 h, samples for each culture were collected and immediately analyzed by flow cytometry. As shown in Fig. 4C, the ubp10 dichlorofluorescein distribution was characterized by a subpopulation of cells (about 25-30%) with high fluorescence, undetectable in the wild type and double mutant strains, in agreement with direct microscopic observation. Similar results have been obtained using dihydrorhodamine 123 (data not shown).
Because the "intensity of respiration" does not induce ROS accumulation (62), the increase in ROS-producing cells suggests that ubp10 mutants have a leaky mitochondrial electron transport. In fact, ROS generation requires an accumulation of reducing equivalents in the middle portion of the electron transfer chain and then a direct one-electron transfer to a O 2 molecule (63). This diversion of the normal electron flow takes place when the respiratory chain is interrupted, for example by cytochrome c release into the cytoplasm. Furthermore, ubp10sir4 mutant did not accumulate ROS indicating that SIR4 disruption was sufficient to abrogate this defect.
ubp10 Cells Display Markers of Apoptosis-An increasing amount of evidence supports the occurrence of apoptosis in S. cerevisiae, a process in which ROS production has been proposed as a prominent regulatory factor (reviewed in Ref. 64). In fact, ROS accumulate after induction of apoptotic death by various stimuli, such as H 2 O 2 and acetic acid (37,65), in old mother cells (62), in cdc13-1 mutants (66), and by ectopic expression of BAX, an apoptotic regulator of BCL-2 family (67). In mammalian cells this group of molecules, including both proapoptoticandantiapoptoticproteins,regulatesthemitochondriadependent cell death process, called "intrinsic pathway," by inducing or preventing release of caspases activators, such as cytochrome c, from mitochondria to the cytosol (68). To investigate if ROS presence in the ubp10 disruptant strain was correlated with the apoptotic phenotype, subcellular markers indicating apoptosis were examined. First, we performed the TUNEL assay, a sensitive tool to detect free 3Ј-OH termini produced by apoptotic DNA cleavage, on ubp10, ubp10sir4, and wild type cells grown on glycerol. As positive control of the TUNEL reaction, an aliquot of the wild type culture was treated with DNase I to produce DNA fragmentation. As shown in Fig. 5A, some ubp10 cells had an intense staining corresponding to the nucleus, comparable with DNase-treated cells and indicative of DNA strand breakage, whereas the double disruptant and the wild type cells displayed a rarely diffuse staining. In parallel, we carried out on the same cultures an annexin V labeling, in order to detect phosphatidylserine exposure to the external layer of the plasma membrane, another hallmark of apoptosis (69,70). Simultaneous incubation with FITC-conjugated annexin V and propidium iodide, a membrane-impermeant fluorochrome, revealed a strong fluorescein green fluorescence in some intact ubp10 cells, whereas all intact spheroplasts of the isogenic control and double mutant showed very slight or no brightness (Fig. 5B). Therefore, in addition to ROS accumulation in the mitochondria, ubp10⌬ mutant displayed a coordinate occurrence of some typical apoptotic features, such as DNA fragmentation and phosphatidylserine exposure, suggesting that loss of the Ubp10 deubiquitinating activity triggers an apoptotic process in an aliquot of the cellular population, which is completely abolished by the lack of Sir4p.
In mammalian cells various substrates of the Ub-dependent proteolytic pathway are involved in the regulation of programmed cell death, but a direct relationship between a deubiquitinating activity and apoptosis was shown only for the tumor suppressor p53. In fact, p53 is a short-lived protein, normally maintained at low level by ubiquitination and consequent proteolysis; deubiquitination by herpes virusassociated ubiquitin-specific protease strongly stabilizes the protein inducing p53-dependent cell growth repression and apoptosis (10). Similarly, the apoptotic phenotype displayed by the ubp10 mutant could be explained by the final consequence of an unbalanced editing of the ubiquitination state of its substrate(s).
The substrates of Ubp10p are so far unknown, but there is evidence to indicate this enzyme acts in the nucleus and is involved in the network of DNA-proteins interactions affecting silencing and the transcriptional regulation of a large number of genes (Ref. 19 and this work). Changes in chromatin structure are linked not only to the activity of the Rap-Sir complexes but also to histone modifications. In this regard, it is worth recalling that histone mono-ubiquitination does not target the protein to degradation, but it is associated with chromatin remodeling (2,71,72). Moreover, in mammalian cells deubiquitination of H2A, coincident with chromatin condensation, occurs in cells undergoing apoptosis as a downstream consequence of caspase activation, but not as a determining apopto-genic stimulus (73). It is conceivable that in the assembly of a repressed chromatin structure both ubiquitination and deubiquitination of the nucleosomal fibers and of the silencing factors are involved. In the ubp10 disruptant changes in the equilibrium between ubiquitination and deubiquitination could affect the assembly or the binding activity of regulatory protein(s); this event might alter local chromatin organization and consequently transcription in agreement with the widely detected remodeling of the transcriptome.