|Home | About | Journals | Submit | Contact Us | Français|
Posttranslational modifications of histone proteins play important roles in the modulation of gene expression. The Saccharomyces cerevisiae (yeast) 2-MDa SAGA (Spt-Ada-Gcn5) complex, a well-studied multisubunit histone modifier, regulates gene expression through Gcn5-mediated histone acetylation and Ubp8-mediated histone deubiquitination. Using a proteomics approach, we determined that the SAGA complex also deubiquitinates nonhistone proteins, including Snf1, an AMP-activated kinase. Ubp8-mediated deubiquitination of Snf1 affects the stability and phosphorylation state of Snf1, thereby affecting Snf1 kinase activity. Others have reported that Gal83 is phosphorylated by Snf1, and we found that deletion of UBP8 causes decreased phosphorylation of Gal83, which is consistent with the effects of Ubp8 loss on Snf1 kinase functions. Overall, our data indicate that SAGA modulates the posttranslational modifications of Snf1 in order to fine-tune gene expression levels.
Ubiquitination of cellular targets regulates many biological processes, from intracellular trafficking to gene expression. Although ubiquitination by ubiquitin (Ub) ligases is widely studied, less is known about subsequent deubiquitination (DUB) of cellular substrates. The identification of genes coding 16 deubiquitinases in the Saccharomyces cerevisiae (yeast) genome and genes coding at least 60 deubiquitinases in the human genome suggests that not only is deubiquitination important but also that deubiquitination may also occur in a substrate-specific manner to regulate specific cellular processes. Despite the lack of knowledge of the targets of many of these deubiquitinases, it is known that some of these enzymes impact cellular growth and function (3, 21, 46). Overexpression of certain deubiquitinases is associated with progression of malignancy in neuroblastomas and a variety of carcinomas, indicating that these enzymes may be oncogenic (27, 31, 38).
USP22, a deubiquitinase associated with the SAGA histone acetyltransferase (HAT) complex, was identified as a member of an 11-gene “death-from-cancer” signature that serves as a predictor of treatment resistance, aggressive growth, and metastasis of human tumors when overexpressed (7, 8). The SAGA complex is highly conserved, and its functions are best characterized in yeast (4, 9, 17, 23, 35). In addition to acetylating histone H3 via the Gcn5 subunit, SAGA also proteolytically cleaves ubiquitin moieties from histone H2B via the Ubp8 subunit, which is an ortholog of USP22 (13, 47). Ubp8-mediated deubiquitination of histone H2B regulates the expression of target genes by modulating the level of histone H3 lysine 4 methylation, a mark that is associated with active transcription (13). In addition, Ubp8 facilitates the recruitment of the C-terminal domain kinase (Ctk1) to target gene promoters via histone H2B deubiquitination, which facilitates the transition from transcription initiation to elongation (43).
Interestingly, recent studies indicate that the functions of USP22 extend beyond histone H2B, as it also deubiquitinates components of the shelterin complex, such as TRF1 (2). Deubiquitination of TRF1 regulates its stability, thereby affecting telomere maintenance. The SAGA complex then may regulate a wide variety of cellular processes that extend beyond alteration of chromatin structures.
To address the possibility that additional factors may be targets of Ubp8/USP22-mediated deubiquitination, we performed a proteomic screen in Saccharomyces cerevisiae. Here, we report that Ubp8 can deubiquitinate Snf1, a highly conserved AMP-activated serine/threonine protein kinase that serves as an energy sensor in the cell. Our data indicate that hyperubiquitination of Snf1 affects its stability as well as its phosphorylation status, which is known to affect Snf1 kinase activity (14, 24, 36). Together, our results suggest that SAGA functions as a master regulator of gene expression not only by altering chromatin structures but also by regulating the stability and activity of transcriptional modulators.
Plasmid pUb221 (a generous gift of Daniel Finley ) contains a 6×His-myc-ubiquitin construct under the control of the copper-inducible CUP1 promoter, TRP1 and URA3 selectable markers, and a 2μm origin of replication. Plasmid pESC-LEU (Stratagene) contains a LEU2 selectable marker, a 2μm origin of replication, and two multicloning sites downstream of the GAL1 and GAL10 yeast promoter. Each multicloning site allows the addition of either a myc or FLAG epitope-tagged gene of interest, respectively. Plasmid pMW10 is a derivative of pESC-LEU and contains a myc-tagged UBP8 gene under the control of the GAL1 promoter. Plasmid pMW18 is a derivative of pESC-LEU and contains a FLAG-tagged SNF1 gene under the control of the GAL10 promoter. Plasmid pMW22 is a derivative of pMW10 and contains a myc-tagged UBP8 gene under the control of the GAL1 promoter and a FLAG-tagged SNF1 gene under the control of the GAL10 promoter. Plasmid pRG145 (a generous gift of Richard Gardner), contains a three-hemagglutinin (3×HA)-tagged ubiquitin construct under the control of the TDH3 promoter, a URA3 selectable marker, and a 2μm origin of replication.
Yeast strains were transformed as previously described (Table 1) (6). To delete UBP8 or SNF1, we PCR amplified the disrupted gene from a yeast deletion strain in the Open Biosystems collection (YSC1021-55154 or YSC1021-555840, respectively) by using primers that annealed 500 bp upstream and downstream of the indicated open reading frame (UBP8 upstream primer oMW3, AACTTTTCCATTTCGGCG; UBP8 downstream primer oMW4, GCCAAAGACGGATATTCTTGG) (SNF1 upstream primer oMW136, AAAAGGATGGGCGTGATGAT; SNF1 downstream primer oMW137, GCAATGGGAGCAAAATTTCC). A one-step gene replacement was performed in the indicated strains by transformation using standard techniques (42). To confirm the deletion of either UBP8 or SNF1, we PCR amplified G418-resistant colonies with primers that annealed within the KANMX6 cassette (KANC, TGATTTTGATGACGAGCGTAAT) and 700 bp downstream of the indicated open reading frame (UBP8 downstream primer oMW140, CCGATGCAGAAAATGAACTCGGTG; SNF1 downstream primer oMW139, CTAACATCTTGTCCAAATGTTGG).
An analysis of Snf1 ubiquitination was performed under denaturing conditions with isogenic wild-type (YSC1178-750067) and ubp8Δ (yMW114) strains expressing a tandem affinity purification (TAP)-tagged version of Snf1 (Snf1-TAP) and copper-inducible 6×His-myc-ubiquitin construct as previously described but with modifications (37). Briefly, overnight cultures grown in 5 ml of synthetic complete medium minus uracil (SC-URA) plus 2% glucose were diluted in 50 ml of SC-URA plus 2% glucose to a starting optical density at 600 nm (OD600) of 0.2 and grown to an OD600 of 0.6 at 30°C. Twenty micromolar Z-Leu-Leu-Leu-al (MG132; Sigma, St. Louis, MO) and 500 μM CuSO4 were then added at final concentrations to each culture for 2 h, and cellular pellets were collected by centrifugation. Samples were lysed in 1 ml of cold buffer A (6 M guanidine-HCl, 0.1 M Na2HPO4/NaH2PO4, 10 mM imidazole, pH to 8.0). Clarified extracts were incubated with 50 μl of preequilibrated nickel-nitrilotriacetic acid (NTA) agarose beads (Qiagen, Valencia, CA) for 2 h and then washed 3 times for 10 min with buffer A. Next, samples were washed 3 times for 10 min in buffer A-buffer TI (1 volume of buffer A, 3 volumes of buffer TI [25 mM Tris-HCl, 20 mM imidazole, pH to 6.8]) and then once with buffer TI and boiled in 5× SDS-PAGE loading buffer containing 200 mM imidazole. Next, eluates were separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with anti-protein A antibodies (Molecular Probes, Eugene, OR) to detect Snf1-TAP.
In order to detect hyperubiquitination of the Snf1-FLAG construct in the absence of UBP8, 50-ml cultures of wild-type and ubp8Δ strains grown in synthetic complete medium minus leucine and uracil (SC-Leu-URA) plus 2% galactose and 2% sucrose in the presence of MG132 (see above) were lysed in 750 μl of IP300 buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 2 mM MgCl2, 0.1% Triton X-100, 10% glycerol, 1 mM β-mercaptoethanol [β-ME], Complete protease inhibitors [Roche, Indianapolis, IN]). Snf1-FLAG was then purified from each lysate by using anti-FLAG agarose beads (Sigma, St. Louis, MO) in the presence of a deubiquitination (DUB) inhibitor (10 mM N-ethylmaleimide [NEM]) and MG132. Eluates were separated by SDS-PAGE and transferred to nitrocellulose membranes which were then probed with either anti-FLAG antibodies (Sigma, St. Louis, MO) or anti-myc antibodies (Santa Cruz Biotechnologies, Santa Cruz, CA) to detect Snf1-FLAG and the 6×His-myc-Ub reporter, respectively.
BY4741 strains containing plasmid pESC-LEU, pMW10, pMW18, or pMW22 grown to a final OD600 of 0.8 in 50 ml of SC-Leu plus 2% galactose and 2% sucrose were lysed in 750 μl of IP300 buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 2 mM MgCl2, 0.1% Triton X-100, 10% glycerol, 1 mM β-ME, Complete protease inhibitors [Roche, Indianapolis, IN]). Lysates were then preincubated with 200 units of DNase I for 1 h followed by a 3-h incubation with either 50 μl of preequilibrated anti-FLAG or anti-myc affinity resin (Sigma, St. Louis, MO) followed by three 10-min washes with 1 ml of IP300. Next, the beads were boiled in 5× SDS-PAGE loading buffer, and the proteins within the supernatant were separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with either anti-FLAG (Sigma, St. Louis, MO) or anti-myc (Santa Cruz Biotechnologies, Santa Cruz, CA) antibodies.
In order to determine if endogenous Snf1 and Ubp8 strains interact, YSC1178-7502243, BY4741, and yMW100 strains were grown to a final OD600 of 0.8 in 50 ml of yeast extract-peptone (YEP) plus 2% glucose and were lysed in 1 ml of IP300 buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 2 mM MgCl2, 0.1% Triton X-100, 10% glycerol, 1 mM β-ME, Complete protease inhibitors [Roche, Indianapolis, IN]). Lysates were then preincubated with 200 units of DNase I for 1 h followed by a 4-h incubation with 40 μl of preequilibrated nickel-NTA agarose beads (Qiagen, Valencia, California) followed by three 10 min-washes with 1 ml of IP300. Next, the beads were boiled in 5× SDS-PAGE loading buffer, and the proteins within the supernatant were separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with either anti-6×His (Qiagen, Valencia, CA) or anti-protein A (Molecular Probes, Eugene, OR) antibodies.
The SAGA complex was purified as previously described with modifications (18). Briefly, overnight cultures of YSC1178-7500046 and yMW122 were diluted into 12 liters of YEP plus 2% glucose to a starting OD600 of 0.2 and grown to a final OD600 of 0.8, collected by centrifugation, and resuspended in 150 ml of SAGA isolation buffer (40 mM HEPES, pH 7.4, 350 mM NaCl, 10% glycerol, 0.1% Tween 20 plus aprotinin [4 μg/ml], leupeptin [4 μg/ml], pepstatin [2 μg/ml], benzamidine [4 μg/ml], and phenylmethylsulfonyl fluoride [PMSF; 200 μg/ml]). Samples were lysed in a bead beater (Biospec) containing 150 ml of cold acid-washed glass beads with five 1-min beating cycles following by 5 min of cooling on ice. Clarified extracts were incubated with 500 μl of preequilibrated IgG sepharose (GE Healthcare, Piscataway, NJ) overnight and then washed twice with 10 ml of SAGA isolation buffer. Next, the beads were washed with 10 ml of TEV cleavage buffer (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% NP-40, 0.5 mM EDTA, 10% glycerol, 1 mM dithiothreitol [DTT] plus aprotinin [4 μg/ml], leupeptin [4 μg/ml], pepstatin [2 μg/ml], benzamidine [4 μg/ml], and phenylmethylsulfonyl fluoride [200 μg/ml]). The beads were then resuspended in 350 μl of TEV cleavage buffer containing 150 U of TEV protease (Invitrogen, Carlsbad, CA) overnight at 4°C. Eluates were collected, flash frozen in liquid nitrogen, and stored at −80°C.
Ubiquitinated H2B purified from yKL142 (kindly provided by Jerry Workman) or ubiquitinated Snf1 purified from yMW16 containing pRG145 was incubated in DUB buffer (50 mM Na-HEPES, pH 7.5, 0.5 mM EDTA, 1 mM DTT, 0.1 M NaCl, 0.1 mg/ml ovalbumin) with SAGA complexes that were isolated from either YSC1178-7500046 or yMW122 strains. The reaction mixtures were incubated for 15 min at 30°C. SDS-PAGE loading buffer (5×) was added to each sample and boiled at 95°C for 5 min. Samples were separated by SDS-PAGE, transferred to nitrocellulose membranes, and subjected to a Western blot analysis with anti-FLAG antibodies (Sigma, St. Louis, MO), anti-HA antibodies (Roche, Indianapolis, IN), and anti-H2B antibodies (Active Motif, Carlsbad, CA). To ensure that equal amounts of SAGA containing and lacking Ubp8 were used in each experiment, each blot was then probed with anti-TAP antibodies (Open Biosystems, Huntsville, AL) to detect Ada2-CBP levels (see Fig. 4B and D, bottom panels).
Recombinant histone H3 (NEB, Ipswich, MA) was incubated for 1 h with SAGA complexes containing or lacking Ubp8 in histone acetyltransferase (HAT) buffer (50 mM Tris-HCl, pH 8.0, 10% glycerol, 0.1 mM EDTA, 1 mM DTT, 1 mM PMSF) containing [3H]acetyl coenzyme A ([3H]acetyl-CoA) at 37°C. Following the incubation, 5× SDS-PAGE loading buffer was added to each sample and samples were boiled at 95°C for 5 min. Samples were separated by SDS-PAGE and stained with Coomassie blue. Next, the gel was incubated with En3Hance (Perkin Elmer Life Sciences, Waltham, MA) for 1 h and washed with distilled water for 30 min. The gel was then dried overnight in a gel dryer on Whatman paper and exposed to autoradiograph film for 24 h to detect [3H]acetyl-histone H3 levels.
Overnight cultures of BY4741 and yMW16 strains were diluted in 5 ml of YEP plus 2% glucose to a starting OD600 of 0.2 and grown to a final OD600 of 0.8 at 30°C. Cell pellets were collected from each sample and stored immediately on dry ice. Next, RNA was isolated from each sample by standard phenol-chloroform extraction procedures. Purified RNA was then subjected to either Northern blotting or quantitative reverse transcription-PCR (RT-PCR) methods.
Overnight cultures of BY4741 and yMW16 strains containing pMW18 were diluted in 50 ml of SC-Leu plus 2% galactose and 2% sucrose to a starting OD600 of 0.2 and grown to a final OD600 of 0.8. At this point, cell pellets were collected by centrifugation and resuspended in 20 ml of SC-Leu without sugar. For time zero, a 1.5-ml aliquot was taken from each sample, and following centrifugation, the remaining cell pellet was immediately placed on dry ice. Next, transcription and translation were terminated by the addition of 4% glucose and 100 μg/ml of cycloheximide (Sigma, St. Louis, MO) to the media. Aliquots of each sample were taken at the indicated time points. Next, whole-cell extracts from each sample were analyzed by Western blotting with anti-FLAG antibodies (Sigma, St. Louis, MO) and anti-Pgk1 antibodies (Molecular Probes, Eugene, OR). Signals were quantified by using ImageQuant software.
In order to determine if the observed proteolysis in ubp8Δ strains was due to proteasome-mediated degradation of Snf1, the experiments were repeated as described above, except that 20 μM Z-Leu-Leu-Leu-al (MG132; Sigma, St. Louis, MO) was added at each step, starting with the media 2 h before time zero.
Overnight cultures of YSC1178-7500067 and yMW114 were diluted in 5 ml of the indicated media to a starting OD600 of 0.2 and grown to a final OD600 of 0.8 at 30°C. Cell pellets were collected by a 30-s centrifugation at 8,000 × g and quickly placed in liquid nitrogen. Cell extracts were then made by standard alkaline lysis and trichloroacetic acid (TCA) precipitation procedures with some modifications (26). Briefly, 2.7 × 107 cells were resuspended in 500 μl of cold double-distilled water (ddH2O). Eighty microliters of NaOH mix (1.85 M NaOH, 7.4% β-ME) was added to each tube, followed by incubation on ice for 10 min. Next, 80 μl of 50% TCA was added to each sample and incubated on ice for 10 min. Each sample was centrifuged for 4 min at 3,300 × g at 4°C. The supernatant was discarded and cell pellets were washed with 500 μl of 1 M Tris-HCl, pH 6.8. Sample pellets were collected and resuspended in 150 μl of 5× SDS-PAGE loading buffer. Samples were then separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with anti-protein A antibodies (Sigma, St. Louis, MO), anti-Pgk1 antibodies (Molecular Probes, Eugene, OR), or anti-phospho-Snf1 T210 antibodies (Cell Signaling, Danvers, MA).
Overnight cultures of YSC1178-7500191, yMW129, and yMW134 grown in YEP plus 2% glucose were diluted in 100 ml of YEP plus 2% glucose to a starting OD600 of 0.2 and grown to a final OD600 of 0.7 at 30°C. Cell pellets were collected from each sample and washed in 50 ml of ddH2O and resuspended in 100 ml of YEP plus 2% glucose, YEP plus 2% galactose, or YEP plus 3% glycerol and 2% ethanol for 45 min. Cell pellets were then collected and frozen in liquid nitrogen. Whole-cell lysates were collected from each strain following resuspension of pellets in 1 ml of IP150 buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM MgCl2, 0.1% Triton X-100, 10% glycerol, 2× Complete protease inhibitors [Roche, Indianapolis, IN], 2× PhosStop phosphatase inhibitors [Roche, Indianapolis, IN]). Lysates were normalized and incubated for 3 h with 20 μl of preequilibrated IgG Sepharose (GE Healthcare, Piscataway, NJ) at 4°C and washed 3 times for 10 min each with IP150. Next, the beads were boiled in 5× SDS-PAGE loading buffer and eluates were separated by SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane. Phospho-Gal83 was detected by staining the membrane with Pro-Q Diamond stain (Invitrogen, Carlsbad, CA). Briefly, the membrane was allowed to completely dry and then dipped in methanol. Next, the membrane was placed into a fix solution (7% acetic acid, 10% methanol) for 10 min and then washed 4 times with ddH2O for 5 min. The membrane was stained for 15 min with Pro-Q Diamond stain and destained with three 5-min washes in destain solution (50 mM sodium acetate [NaOAc], pH 4.0, 20% acetonitrile). The membrane was then dried and phospho-Gal83 was visualized on a Storm840 scanner. Total Gal83 was detected with an anti-protein A antibody (Sigma, St. Louis, MO) and visualized by using the Storm840 scanner.
To identify potential Ubp8 substrates, we isolated ubiquitinated proteins from wild-type and ubp8Δ strains. Our reasoning was that ubiquitination of Ubp8 substrates would be increased in its absence, leading to an increased abundance of these proteins in ubiquitin isolations from ubp8Δ strains. The details of the isolation and a full list of potential targets identified by this screen will be presented elsewhere, but mass spectrometry of the isolated proteins identified Snf1 as a potential target of Ubp8. Snf1 was identified in two independent experiments, as was the known Ubp8 target, histone H2B (data not shown).
Snf1 is the founding member of the SNF1 family of AMP kinases (AMPKs), which is highly conserved from yeast to mammals (5, 10, 30). While it is involved in a wide variety of cellular processes, its major function in yeast is to induce expression of glucose-repressed genes when cells are grown on nonfermentable carbon sources through phosphorylation-mediated activation and deactivation of transcriptional regulators (28). To confirm the results of our biochemical screen, we purified 6×His-myc-ubiquitin from wild-type and ubp8Δ strains expressing a tandem affinity purification (TAP)-tagged version of Snf1 and a 6×His-myc epitope-tagged ubiquitin construct. After extensive washing, Ni-NTA-bound proteins were resolved by SDS-PAGE and analyzed by Western blot analyses with anti-TAP antibodies (Fig. 1 A and B). We detected some ubiquitination of Snf1 in wild-type strains, but the abundance of the Snf1 ubiquitination was notably increased upon loss of Ubp8, further indicating that Snf1 is a target of Ubp8-mediated deubiquitination.
Ubiquitination of a protein can trigger its proteolysis. The isolations described above were performed in the presence of proteasome inhibitors in order to capture and enrich samples for ubiquitinated Ubp8 targets. To determine whether degradation of Snf1 is increased upon its hyperubiquitination, we determined whether steady-state levels of Snf1 were decreased in ubp8Δ strains. Importantly, we found that SNF1 mRNA levels were unchanged in wild-type and ubp8Δ strains (Fig. 1C). Western blots, however, revealed that Snf1 protein levels decreased in the absence of Ubp8 (Fig. 2 A and B).
To confirm that the decrease in Snf1 levels occurs posttranslationally, we monitored the stability of Snf1 expressed under the control of a galactose-inducible promoter at various time points following a shift of cells to repressing conditions (the addition of glucose) in the presence of cycloheximide, thus blocking transcription and translation of Snf1. Snf1 proteolysis occurred rapidly in ubp8Δ strains relative to wild-type strains (Fig. 2C and D). Further, this decrease in Snf1 stability was dependent on the carbon source, as it was observed only for ubp8Δ strains grown in galactose (Fig. 2E). Together, these data suggest that Ubp8 regulates Snf1 levels posttranslationally during growth on alternative carbon sources. In addition, the MG132 proteasome inhibitor stabilized Snf1 in ubp8Δ strains (ubp8Δ plus MG132; Fig. 2C and D), indicating that Ubp8-mediated deubiquitination protects Snf1 from proteasome-mediated degradation.
The finding that the ubiquitination status and stability of Snf1 are altered in ubp8Δ strains strongly suggests that Snf1 is a novel nonhistone substrate of Ubp8. If Ubp8 deubiquitinates Snf1, then we would expect to find interactions between these two proteins in vivo. To test this hypothesis, we coexpressed a FLAG epitope-tagged form of Snf1 and a myc epitope-tagged form of Ubp8 expressed from plasmid constructs. Western blots of anti-myc or anti-FLAG immunoprecipitates were probed with anti-FLAG or anti-myc antibodies to confirm that Snf1 and Ubp8 interact in vivo (Fig. 3 A and B). In addition, we were still able to detect these interactions in the presence of DNase and with the endogenously tagged proteins, indicating that the interactions are not mediated by DNA or are an artifact of the overexpression of these proteins, further supporting our hypothesis that Snf1 is a substrate of Ubp8 (Fig. 3A, B, and C).
To determine if Ubp8 can deubiquitinate Snf1 in vitro, we purified SAGA from strains that expressed a TAP-tagged version of Ada2, a component of the SAGA complex. In addition, we purified FLAG-tagged Snf1 from a strain that also expressed HA-tagged ubiquitin to be used as a substrate in these assays (Fig. 4 C). For controls, in addition to purifying the known Ubp8 target, histone H2B, from a strain that also expressed HA-tagged ubiquitin (Fig. 4A), we performed an in vitro DUB assay with Ub-AMC (7-amino-4-methylcoumarin) to monitor the DUB activity of the purified complexes (Fig. 4F). Ubiquitinated Snf1 or H2B was incubated in the absence and presence of the SAGA complex and analyzed by use of anti-FLAG, anti-H2B, and anti-HA Western blots to determine whether SAGA could cleave ubiquitin from these proteins (Fig. 4B and D). SAGA cleaved ubiquitin from both histone H2B (Fig. 4B, lane 2) and Snf1 (Fig. 4D, lane 2). To confirm that the ability of SAGA to deubiquitinate Snf1 was specific to the activity of Ubp8, we also purified the SAGA complex from ubp8Δ strains. While the in vitro histone acetyltransferase activity of SAGA remained intact in the ubp8Δ strain (Fig. 4E, bottom panel, compare lanes 3 and 4), we observed a complete loss of SAGA-mediated deubiquitination of both histone H2B and Snf1, indicating that the observed deubiquitination described above was mediated by Ubp8 (Fig. 4B and D, lanes 3, and F). Together, these data indicate that Ubp8 can directly deubiquitinate Snf1.
The activity of Snf1 and related AMP kinases is highly regulated (34). In mammals, AMPK acts as an energy sensor that becomes activated when cellular ratios of AMP to ATP are altered, which likely contributes to survival of a variety of tumors in nutrient-poor or hypoxic conditions. For example, for human glioblastoma multiform tumors, microenvironments of hypoxia that alter the energy balance of AMP and ATP result in increased expression of AMPK. Further, in glioma cell lines, hyperactive AMPK causes an increase in the expression of vascular endothelial growth factor (VEGF), a factor necessary for angiogenesis and survival of these high-grade tumors (25).
The activity of the Snf1 family is regulated in many different ways. One highly conserved mechanism involves the phosphorylation of the T loop at threonine 210 (T210) (14, 19, 24, 36), which closely parallels the activation of Snf1 and many other AMP kinases. When yeast cells grown in glucose are switched to less-preferred carbon sources, Snf1 becomes phosphorylated and then activates the expression of genes involved in utilization of these carbon sources by regulating the activity of downstream transcriptional modulators (12, 20, 28, 29, 33, 39). Interestingly, hyperubiquitination of mammalian Snf1-related AMP-activated protein kinases is associated with a marked decrease in T-loop phosphorylation, suggesting that ubiquitination regulates their activity (1).
To determine whether the hyperubiquitination of Snf1 that we observed for ubp8Δ strains affects its phosphorylation status, we monitored Snf1 T210 phosphorylation in wild-type and ubp8Δ strains grown in activating (galactose) and nonactivating (glucose) carbon sources. Since Snf1 steady-state levels are decreased in ubp8Δ strains (Fig. 2A), we normalized the protein loads of Snf1 for these analyses. As expected, Snf1 T210 phosphorylation was regulated by the carbon source (Fig. 5 B [top panel, compare lanes 1 and 3] and C), as it was increased in galactose-grown cells. In contrast, Snf1 T210 phosphorylation was reduced in ubp8Δ strains grown in galactose (Fig. 5B [compare lanes 3 and 4] and C), suggesting that, as for other AMP kinases, hyperubiquitination of Snf1 in the absence of Ubp8 affects T210 phosphorylation and potentially Snf1 activation.
Decreased Snf1 T210 phosphorylation upon increased ubiquitination could lead to a decrease in Snf1 kinase activity. If so, then loss of Ubp8 might cause changes in gene expression similar to those caused by loss of Snf1. Therefore, we examined previously published microarray data and found significant overlaps between changes in gene expression reported for snf1Δ and ubp8Δ strains (16, 45). However, it is difficult to distinguish which changes are direct or indirect, as the SAGA complex and Snf1 each regulate many genes (Table 2) . However, previously published chromatin immunoprecipitation microarray analysis (ChIP-chip) data revealed significant overlaps in Snf1 and Ubp8 binding at upstream activation sites (UAS) and transcriptional start sites (TSS) of many genes whose expression was affected by loss of these factors. This overlap in function at these genes supports the idea that Ubp8 modulates the activity of Snf1 to fine-tune the regulation of Snf1 target genes (Fig. 5A and Table 2) (16, 40, 45).
To more directly determine whether Snf1 activity is affected by Ubp8 loss, we examined the phosphorylation status of a Snf1 target protein. The Snf1 kinase complex is composed of a catalytic alpha subunit (Snf1), one of three beta subunits (Gal83, Sip1, and Sip2), and a gamma subunit (Snf4) (11). Each of the three beta subunits interacts with Snf1 and Snf4 to form three distinct isoforms of the Snf1 complex that have overlapping functions. Strains containing null mutations in any one of these proteins are viable on all carbon sources and show no differences in the phosphorylation status of downstream protein targets (32). Conversely, strains containing null mutations in all three of the beta subunits of the Snf1 complex phenocopy the growth and activity defects observed in snf1 null mutants when grown on nonfermentable carbon sources (32). While the beta subunits appear to have shared functions, they can also confer substrate specificity to the Snf1 kinase complex, and they have distinct roles in the cell during growth on specific carbon sources (32).
The Snf1 complex containing the Gal83 beta subunit localizes to the nucleus during a switch from high- to low-glucose-containing media. Therefore, it is likely that this isoform of the Snf1 complex is posttranslationally modified by Ubp8. In addition, Gal83 is phosphorylated by Snf1 in vivo and in vitro (22, 41). Therefore, to determine if the hypophosphorylation of Snf1 that we observe in ubp8Δ strains corresponds to a decrease in Snf1 activity, we monitored the phosphorylation status of Gal83 in wild-type, ubp8Δ, and snf1Δ strains grown in activating (galactose) and nonactivating (glucose) carbon sources. Like Snf1 T210 phosphorylation, the phosphorylation of Gal83 is regulated by the carbon source (Fig. 5D, top panel, compare lanes 1 and 4), as it is phosphorylated when cells are grown in activating conditions. However, Gal83 phosphorylation is decreased in ubp8Δ strains (Fig. 5D, top panel, compare lanes 1 and 2), suggesting that the decrease in Snf1 stability and T210 phosphorylation that is observed in these strains corresponds to a decrease in Snf1 activity.
One of the most studied modes of Snf1 regulation is the phosphorylation of threonine 210, a mark which closely follows Snf1 activation (14, 24, 36). In this study, we found that in addition to being modified by phosphorylation, Snf1 is modified by ubiquitination. Moreover, we discovered that Snf1 ubiquitination is regulated by the SAGA complex, as Snf1 becomes hyperubiquitinated in the absence of Ubp8 (Fig. 1 and and6).6). Snf1 physically interacts with Ubp8 in vivo, and the SAGA complex deubiquitinates Snf1 in vitro, in a Ubp8-dependent manner. To our knowledge, Snf1 is the first nonhistone substrate defined for the yeast DUB module of SAGA.
What is the biological function of Snf1 ubiquitination? One method to determine whether hyperubiquitination of Snf1 in ubp8Δ strains affects Snf1 activity is to determine whether snf1Δ and ubp8Δ strains share growth defects. However, one problem with testing this hypothesis is that ubp8Δ strains, unlike snf1Δ strains, have growth defects on preferred (control) carbon sources (i.e., glucose [data not shown]), which makes interpretation of any results difficult. However, our data indicate that Ubp8 does in fact regulate Snf1, as it regulates both the half-life of Snf1 and its kinase activity. Interestingly, while ubiquitination of other AMP kinases (AMPKs) has been described, only a few studies address the regulation of these enzymes by ubiquitination. The mammalian USP9X deubiquitinase regulates the ubiquitination status of the MARK4 and NUAK1 AMPK-related kinases (1). Similar to what was seen in our findings, MARK4 and NUAK1 are hyperubiquitinated in the absence of USP9X. However, hyperubiquitination of these enzymes does not seem to affect their stability but rather has a negative effect on MARK4 and NUAK1 T-loop phosphorylation. Since ubiquitination of Snf1 not only affected its stability, but like MARK4 and NUAK1, also had a negative effect on its phosphorylation status, the activity of Snf1 is likely tightly regulated by the opposing actions of Ubp8 and ubiquitin ligases. In yeast and mammals, the major upstream activating kinases of the SNF1 family of proteins (in yeast, Sak1, Tos3, and Elm1, and in mammals, LKB1 and CaMKK) are constitutively active (15). However, when cells are grown in activating conditions, only a fraction of these kinase targets is phosphorylated. Therefore, it is possible that in a system that is “always on,” the ubiquitination/deubiquitination cycles of SNF1 family members have evolved to allow only a portion of each family member to be recognized by or accessible to these upstream kinases. Similarly, Ubp8-mediated deubiquitination of Snf1 may have evolved to fine-tune the spatial and temporal activation of Snf1.
Snf1 can directly phosphorylate histone proteins, transcriptional modulators, and the beta subunits of the Snf1 kinase complex isoforms. Mutations that inactivate Snf1 cause a decrease in the phosphorylation of these downstream targets (22, 41). We show that SAGA, through the Ubp8 deubiquitinase, fine-tunes these activities, as hyperubiquitination of Snf1 decreases phosphorylation of both Snf1 and its downstream target, Gal83. Therefore, Ubp8 functions positively reinforce Snf1-mediated activities in the cell (Fig. 6).
Our data not only provide additional information about the molecular mechanisms of the SAGA complex but also suggest that ubiquitination of AMPKs provides a mechanism that is conserved from yeast to humans for the regulation of their activity. Future work will identify the ubiquitin ligase that is responsible for Snf1 ubiquitination and the mechanism by which Snf1 ubiquitination inhibits its activation.
Our current and previous works demonstrate that SAGA utilizes multiple mechanisms to fine-tune gene expression beyond chromatin structure. Therefore, like for the Snf1 AMPK, the identification and characterization of these additional targets of the SAGA complex will be critical to understanding the role that USP22 and SAGA play in the death-from-cancer signature that is observed in various carcinomas (7, 8).
We thank William Dubinsky (The University of Texas Dental Branch) for performing the mass spectrometry experiments. We thank Ambro van Hoof, Daneen Grossman, Kevin Morano, and Hugo Tapia (The University of Texas Health Science Center at Houston), Hugo Bellen and Nikos Giagtzoglou (Baylor College of Medicine), and Jerry Workman and Kenneth Lee (Stowers Institute) for yeast strains, reagents, and equipment that were used to conduct this work. We also thank Daniel Finley (Harvard Medical School) and Richard Gardner (University of Washington) for generously sharing plasmids. We thank Ambro van Hoof, John Latham, Yi Chun Chen, Jill Butler, Boyko Atanassov, and Marek Napierela (The University of Texas M. D. Anderson Cancer Center) for fruitful discussions.
This work was supported by the National Institute of Child Health and Human Development Training Grant in “Differentiation and Development” to M.A.W. (2 T32 HD07325) and grants from the NIH (R01GM51189) and the MDACC Senior Research Trust to S.Y.R.D.
Published ahead of print on 31 May 2011.